Main Page | Modules | Namespace List | Class Hierarchy | Alphabetical List | Class List | Directories | File List | Namespace Members | Class Members | File Members

emdata.cpp

Go to the documentation of this file.
00001 
00005 /*
00006  * Author: Steven Ludtke, 04/10/2003 (sludtke@bcm.edu)
00007  * Copyright (c) 2000-2006 Baylor College of Medicine
00008  *
00009  * This software is issued under a joint BSD/GNU license. You may use the
00010  * source code in this file under either license. However, note that the
00011  * complete EMAN2 and SPARX software packages have some GPL dependencies,
00012  * so you are responsible for compliance with the licenses of these packages
00013  * if you opt to use BSD licensing. The warranty disclaimer below holds
00014  * in either instance.
00015  *
00016  * This complete copyright notice must be included in any revised version of the
00017  * source code. Additional authorship citations may be added, but existing
00018  * author citations must be preserved.
00019  *
00020  * This program is free software; you can redistribute it and/or modify
00021  * it under the terms of the GNU General Public License as published by
00022  * the Free Software Foundation; either version 2 of the License, or
00023  * (at your option) any later version.
00024  *
00025  * This program is distributed in the hope that it will be useful,
00026  * but WITHOUT ANY WARRANTY; without even the implied warranty of
00027  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
00028  * GNU General Public License for more details.
00029  *
00030  * You should have received a copy of the GNU General Public License
00031  * along with this program; if not, write to the Free Software
00032  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
00033  *
00034  * */
00035 
00036 #include "emdata.h"
00037 #include "all_imageio.h"
00038 #include "ctf.h"
00039 #include "processor.h"
00040 #include "aligner.h"
00041 #include "cmp.h"
00042 #include "emfft.h"
00043 #include "projector.h"
00044 #include "geometry.h"
00045 
00046 #include <gsl/gsl_sf_bessel.h>
00047 #include <gsl/gsl_errno.h>
00048 
00049 #include <iomanip>
00050 #include <complex>
00051 
00052 #include <algorithm> // fill
00053 #include <cmath>
00054 
00055 #ifdef WIN32
00056         #define M_PI 3.14159265358979323846f
00057 #endif  //WIN32
00058 
00059 #define EMDATA_EMAN2_DEBUG 0
00060 
00061 #ifdef EMAN2_USING_CUDA
00062 //#include <driver_functions.h>
00063 #include "cuda/cuda_processor.h"
00064 #include "cuda/cuda_emfft.h"
00065 #endif // EMAN2_USING_CUDA
00066 
00067 using namespace EMAN;
00068 using namespace std;
00069 using namespace boost;
00070 
00071 int EMData::totalalloc=0;               // mainly used for debugging/memory leak purposes
00072 
00073 EMData::EMData() :
00074 #ifdef EMAN2_USING_CUDA
00075                 cudarwdata(0), cudarodata(0), num_bytes(0), nextlistitem(0), prevlistitem(0), roneedsupdate(0), 
00076 #endif //EMAN2_USING_CUDA
00077                 attr_dict(), rdata(0), supp(0), flags(0), changecount(0), nx(0), ny(0), nz(0), nxy(0), nxyz(0), xoff(0), yoff(0),
00078                 zoff(0), all_translation(),     path(""), pathnum(0), rot_fp(0)
00079 
00080 {
00081         ENTERFUNC;
00082 
00083         attr_dict["apix_x"] = 1.0f;
00084         attr_dict["apix_y"] = 1.0f;
00085         attr_dict["apix_z"] = 1.0f;
00086 
00087         attr_dict["is_complex"] = int(0);
00088         attr_dict["is_complex_x"] = int(0);
00089         attr_dict["is_complex_ri"] = int(1);
00090 
00091         attr_dict["datatype"] = (int)EMUtil::EM_FLOAT;
00092 
00093         EMData::totalalloc++;
00094 #ifdef MEMDEBUG2
00095         printf("EMDATA+  %4d    %p\n",EMData::totalalloc,this);
00096 #endif
00097 
00098         EXITFUNC;
00099 }
00100 
00101 EMData::EMData(const string& filename, int image_index) :
00102 #ifdef EMAN2_USING_CUDA
00103                 cudarwdata(0), cudarodata(0), num_bytes(0), nextlistitem(0), prevlistitem(0), roneedsupdate(0),
00104 #endif //EMAN2_USING_CUDA
00105                 attr_dict(), rdata(0), supp(0), flags(0), changecount(0), nx(0), ny(0), nz(0), nxy(0), nxyz(0), xoff(0), yoff(0), zoff(0),
00106                 all_translation(),      path(filename), pathnum(image_index), rot_fp(0)
00107 {
00108         ENTERFUNC;
00109 
00110         attr_dict["apix_x"] = 1.0f;
00111         attr_dict["apix_y"] = 1.0f;
00112         attr_dict["apix_z"] = 1.0f;
00113 
00114         attr_dict["is_complex"] = int(0);
00115         attr_dict["is_complex_x"] = int(0);
00116         attr_dict["is_complex_ri"] = int(1);
00117 
00118         attr_dict["datatype"] = (int)EMUtil::EM_FLOAT;
00119 
00120         this->read_image(filename, image_index);
00121 
00122         update();
00123         EMData::totalalloc++;
00124 #ifdef MEMDEBUG2
00125         printf("EMDATA+  %4d    %p\n",EMData::totalalloc,this);
00126 #endif
00127 
00128         EXITFUNC;
00129 }
00130 
00131 EMData::EMData(const EMData& that) :
00132 #ifdef EMAN2_USING_CUDA
00133                 cudarwdata(0), cudarodata(0), num_bytes(0), nextlistitem(0), prevlistitem(0), roneedsupdate(0),
00134 #endif //EMAN2_USING_CUDA
00135                 attr_dict(that.attr_dict), rdata(0), supp(0), flags(that.flags), changecount(that.changecount), nx(that.nx), ny(that.ny), nz(that.nz),
00136                 nxy(that.nx*that.ny), nxyz((size_t)that.nx*that.ny*that.nz), xoff(that.xoff), yoff(that.yoff), zoff(that.zoff),all_translation(that.all_translation),   path(that.path),
00137                 pathnum(that.pathnum), rot_fp(0)
00138 {
00139         ENTERFUNC;
00140         
00141         float* data = that.rdata;
00142         size_t num_bytes = (size_t)nx*ny*nz*sizeof(float);
00143         if (data && num_bytes != 0)
00144         {
00145                 rdata = (float*)EMUtil::em_malloc(num_bytes);
00146                 EMUtil::em_memcpy(rdata, data, num_bytes);
00147         }
00148 #ifdef EMAN2_USING_CUDA
00149         if (num_bytes != 0 && that.cudarwdata != 0) {
00150                 //cout << "That copy constructor" << endl;
00151                 rw_alloc();
00152                 cudaError_t error = cudaMemcpy(cudarwdata,that.cudarwdata,num_bytes,cudaMemcpyDeviceToDevice);
00153                 if ( error != cudaSuccess ) throw UnexpectedBehaviorException("cudaMemcpy failed in EMData copy construction with error: " + string(cudaGetErrorString(error)));
00154         }
00155 #endif //EMAN2_USING_CUDA
00156 
00157         if (that.rot_fp != 0) rot_fp = new EMData(*(that.rot_fp));
00158 
00159         EMData::totalalloc++;
00160 #ifdef MEMDEBUG2
00161         printf("EMDATA+  %4d    %p\n",EMData::totalalloc,this);
00162 #endif
00163 
00164         ENTERFUNC;
00165 }
00166 
00167 EMData& EMData::operator=(const EMData& that)
00168 {
00169         ENTERFUNC;
00170 
00171         if ( this != &that )
00172         {
00173                 free_memory(); // Free memory sets nx,ny and nz to 0
00174 
00175                 // Only copy the rdata if it exists, we could be in a scenario where only the header has been read
00176                 float* data = that.rdata;
00177                 size_t num_bytes = that.nx*that.ny*that.nz*sizeof(float);
00178                 if (data && num_bytes != 0)
00179                 {
00180                         nx = 1; // This prevents a memset in set_size
00181                         set_size(that.nx,that.ny,that.nz);
00182                         EMUtil::em_memcpy(rdata, data, num_bytes);
00183                 }
00184 
00185                 flags = that.flags;
00186 
00187                 all_translation = that.all_translation;
00188 
00189                 path = that.path;
00190                 pathnum = that.pathnum;
00191                 attr_dict = that.attr_dict;
00192 
00193                 xoff = that.xoff;
00194                 yoff = that.yoff;
00195                 zoff = that.zoff;
00196 
00197 #ifdef EMAN2_USING_CUDA
00198                 cout << "That copy constructor #2" << endl;
00199                 if (num_bytes != 0 && that.cudarwdata != 0) {
00200                         rw_alloc();
00201                         cudaError_t error = cudaMemcpy(cudarwdata,that.cudarwdata,num_bytes,cudaMemcpyDeviceToDevice);
00202                         if ( error != cudaSuccess ) throw UnexpectedBehaviorException("cudaMemcpy failed in EMData copy construction with error: " + string(cudaGetErrorString(error)));
00203                         
00204                 }
00205 #endif //EMAN2_USING_CUDA
00206 
00207                 changecount = that.changecount;
00208 
00209                 if (that.rot_fp != 0) rot_fp = new EMData(*(that.rot_fp));
00210                 else rot_fp = 0;
00211         }
00212         EXITFUNC;
00213         return *this;
00214 }
00215 
00216 EMData::EMData(int nx, int ny, int nz, bool is_real) :
00217 #ifdef EMAN2_USING_CUDA
00218                 cudarwdata(0), cudarodata(0), num_bytes(0), nextlistitem(0), prevlistitem(0), roneedsupdate(0),
00219 #endif //EMAN2_USING_CUDA
00220                 attr_dict(), rdata(0), supp(0), flags(0), changecount(0), nx(0), ny(0), nz(0), nxy(0), nxyz(0), xoff(0), yoff(0), zoff(0),
00221                 all_translation(),      path(""), pathnum(0), rot_fp(0)
00222 {
00223         ENTERFUNC;
00224 
00225         // used to replace cube 'pixel'
00226         attr_dict["apix_x"] = 1.0f;
00227         attr_dict["apix_y"] = 1.0f;
00228         attr_dict["apix_z"] = 1.0f;
00229 
00230         if(is_real) {   // create a real image [nx, ny, nz]
00231                 attr_dict["is_complex"] = int(0);
00232                 attr_dict["is_complex_x"] = int(0);
00233                 attr_dict["is_complex_ri"] = int(1);
00234                 set_size(nx, ny, nz);
00235         }
00236         else {  //create a complex image which real dimension is [ny, ny, nz]
00237                 int new_nx = nx + 2 - nx%2;
00238                 set_size(new_nx, ny, nz);
00239 
00240                 attr_dict["is_complex"] = int(1);
00241 
00242                 if(ny==1 && nz ==1)     {
00243                         attr_dict["is_complex_x"] = int(1);
00244                 }
00245                 else {
00246                         attr_dict["is_complex_x"] = int(0);
00247                 }
00248 
00249                 attr_dict["is_complex_ri"] = int(1);
00250                 attr_dict["is_fftpad"] = int(1);
00251 
00252                 if(nx%2 == 1) {
00253                         attr_dict["is_fftodd"] = 1;
00254                 }
00255         }
00256 
00257         this->to_zero();
00258         update();
00259         EMData::totalalloc++;
00260 #ifdef MEMDEBUG2
00261         printf("EMDATA+  %4d    %p\n",EMData::totalalloc,this);
00262 #endif
00263 
00264         EXITFUNC;
00265 }
00266 
00267 
00268 EMData::EMData(float* data, const int x, const int y, const int z, const Dict& attr_dict) :
00269 #ifdef EMAN2_USING_CUDA
00270                 cudarwdata(0), cudarodata(0), num_bytes(0), nextlistitem(0), prevlistitem(0), roneedsupdate(0),
00271 #endif //EMAN2_USING_CUDA
00272                 attr_dict(attr_dict), rdata(data), supp(0), flags(0), changecount(0), nx(x), ny(y), nz(z), nxy(x*y), nxyz((size_t)x*y*z), xoff(0),
00273                 yoff(0), zoff(0), all_translation(), path(""), pathnum(0), rot_fp(0)
00274 {
00275         ENTERFUNC;
00276         // used to replace cube 'pixel'
00277         attr_dict["apix_x"] = 1.0f;
00278         attr_dict["apix_y"] = 1.0f;
00279         attr_dict["apix_z"] = 1.0f;
00280 
00281         EMData::totalalloc++;
00282 #ifdef MEMDEBUG2
00283         printf("EMDATA+  %4d    %p\n",EMData::totalalloc,this);
00284 #endif
00285 
00286         update();
00287         EXITFUNC;
00288 }
00289 
00290 #ifdef EMAN2_USING_CUDA
00291 
00292 EMData::EMData(float* data, float* cudadata, const int x, const int y, const int z, const Dict& attr_dict) :
00293                 cudarwdata(cudadata), cudarodata(0), num_bytes(x*y*z*sizeof(float)), nextlistitem(0), prevlistitem(0), roneedsupdate(0),
00294                 attr_dict(attr_dict), rdata(data), supp(0), flags(0), changecount(0), nx(x), ny(y), nz(z), nxy(x*y), nxyz((size_t)x*y*z), xoff(0),
00295                 yoff(0), zoff(0), all_translation(), path(""), pathnum(0), rot_fp(0)
00296 {
00297         ENTERFUNC;
00298 
00299         // used to replace cube 'pixel'
00300         attr_dict["apix_x"] = 1.0f;
00301         attr_dict["apix_y"] = 1.0f;
00302         attr_dict["apix_z"] = 1.0f;
00303 
00304         EMData::totalalloc++;
00305 #ifdef MEMDEBUG2
00306         printf("EMDATA+  %4d    %p\n",EMData::totalalloc,this);
00307 #endif
00308 
00309         update();
00310         EXITFUNC;
00311 }
00312 
00313 #endif //EMAN2_USING_CUDA
00314 
00315 //debug
00316 using std::cout;
00317 using std::endl;
00318 EMData::~EMData()
00319 {
00320         ENTERFUNC;
00321         free_memory();
00322 
00323 #ifdef EMAN2_USING_CUDA
00324         if(cudarwdata){rw_free();}
00325         if(cudarodata){ro_free();}
00326 #endif // EMAN2_USING_CUDA
00327         EMData::totalalloc--;
00328 #ifdef MEMDEBUG2
00329         printf("EMDATA-  %4d    %p\n",EMData::totalalloc,this);
00330 #endif
00331         EXITFUNC;
00332 }
00333 
00334 void EMData::clip_inplace(const Region & area,const float& fill_value)
00335 {
00336         // Added by d.woolford
00337         ENTERFUNC;
00338 
00339         // Store the current dimension values
00340         int prev_nx = nx, prev_ny = ny, prev_nz = nz;
00341         size_t prev_size = (size_t)nx*ny*nz;
00342 
00343         // Get the zsize, ysize and xsize of the final area, these are the new dimension sizes of the pixel data
00344         int new_nz = ( area.size[2]==0 ? 1 : (int)area.size[2]);
00345         int new_ny = ( area.size[1]==0 ? 1 : (int)area.size[1]);
00346         int new_nx = (int)area.size[0];
00347 
00348         if ( new_nz < 0 || new_ny < 0 || new_nx < 0 )
00349         {
00350                 // Negative image dimensions were never tested nor considered when creating this implementation
00351                 throw ImageDimensionException("New image dimensions are negative - this is not supported in the clip_inplace operation");
00352         }
00353 
00354         size_t new_size = (size_t)new_nz*new_ny*new_nx;
00355 
00356         // Get the translation values, they are used to construct the ClipInplaceVariables object
00357         int x0 = (int) area.origin[0];
00358         int y0 = (int) area.origin[1];
00359         int z0 = (int) area.origin[2];
00360 
00361         // Get a object that calculates all the interesting variables associated with the clip inplace operation
00362         ClipInplaceVariables civ(prev_nx, prev_ny, prev_nz, new_nx, new_ny, new_nz, x0, y0, z0);
00363 
00364         get_data(); // Do this here to make sure rdata is up to date, applicable if GPU stuff is occuring
00365         // Double check to see if any memory shifting even has to occur
00366         if ( x0 > prev_nx || y0 > prev_ny || z0 > prev_nz || civ.x_iter == 0 || civ.y_iter == 0 || civ.z_iter == 0)
00367         {
00368                 // In this case the volume has been shifted beyond the location of the pixel rdata and
00369                 // the client should expect to see a volume with nothing in it.
00370 
00371                 // Set size calls realloc,
00372                 set_size(new_nx, new_ny, new_nz);
00373 
00374                 // Set pixel memory to zero - the client should expect to see nothing
00375                 EMUtil::em_memset(rdata, 0, (size_t)new_nx*new_ny*new_nz);
00376 
00377                 return;
00378         }
00379 
00380         // Resize the volume before memory shifting occurs if the new volume is larger than the previous volume
00381         // All of the pixel rdata is guaranteed to be at the start of the new volume because realloc (called in set size)
00382         // guarantees this.
00383         if ( new_size > prev_size )
00384                 set_size(new_nx, new_ny, new_nz);
00385 
00386         // Store the clipped row size.
00387         size_t clipped_row_size = (civ.x_iter) * sizeof(float);
00388 
00389         // Get the new sector sizes to save multiplication later.
00390         size_t new_sec_size = new_nx * new_ny;
00391         size_t prev_sec_size = prev_nx * prev_ny;
00392 
00393         // Determine the memory locations of the source and destination pixels - at the point nearest
00394         // to the beginning of the volume (rdata)
00395         size_t src_it_begin = civ.prv_z_bottom*prev_sec_size + civ.prv_y_front*prev_nx + civ.prv_x_left;
00396         size_t dst_it_begin = civ.new_z_bottom*new_sec_size + civ.new_y_front*new_nx + civ.new_x_left;
00397 
00398         // This loop is in the forward direction (starting at points nearest to the beginning of the volume)
00399         // it copies memory only when the destination pointer is less the source pointer - therefore
00400         // ensuring that no memory "copied to" is yet to be "copied from"
00401         for (int i = 0; i < civ.z_iter; ++i) {
00402                 for (int j = 0; j < civ.y_iter; ++j) {
00403 
00404                         // Determine the memory increments as dependent on i and j
00405                         // This could be optimized so that not so many multiplications are occurring...
00406                         size_t dst_inc = dst_it_begin + j*new_nx + i*new_sec_size;
00407                         size_t src_inc = src_it_begin + j*prev_nx + i*prev_sec_size;
00408                         float* local_dst = rdata + dst_inc;
00409                         float* local_src = rdata + src_inc;
00410 
00411                         if ( dst_inc >= src_inc )
00412                         {
00413                                 // this is fine, it will happen now and then and it will be necessary to continue.
00414                                 // the tempatation is to break, but you can't do that (because the point where memory intersects
00415                                 // could be in this slice - and yes, this aspect could be optimized).
00416                                 continue;
00417                         }
00418 
00419                         // Asserts are compiled only in debug mode
00420                         // This situation not encountered in testing thus far
00421                         Assert( dst_inc < new_size && src_inc < prev_size && dst_inc >= 0 && src_inc >= 0 );
00422 
00423                         // Finally copy the memory
00424                         EMUtil::em_memcpy(local_dst, local_src, clipped_row_size);
00425                 }
00426         }
00427 
00428         // Determine the memory locations of the source and destination pixels - at the point nearest
00429         // to the end of the volume (rdata+new_size)
00430         size_t src_it_end = prev_size - civ.prv_z_top*prev_sec_size - civ.prv_y_back*prev_nx - prev_nx + civ.prv_x_left;
00431         size_t dst_it_end = new_size - civ.new_z_top*new_sec_size - civ.new_y_back*new_nx - new_nx + civ.new_x_left;
00432 
00433         // This loop is in the reverse direction (starting at points nearest to the end of the volume).
00434         // It copies memory only when the destination pointer is greater than  the source pointer therefore
00435         // ensuring that no memory "copied to" is yet to be "copied from"
00436         for (int i = 0; i < civ.z_iter; ++i) {
00437                 for (int j = 0; j < civ.y_iter; ++j) {
00438 
00439                         // Determine the memory increments as dependent on i and j
00440                         size_t dst_inc = dst_it_end - j*new_nx - i*new_sec_size;
00441                         size_t src_inc = src_it_end - j*prev_nx - i*prev_sec_size;
00442                         float* local_dst = rdata + dst_inc;
00443                         float* local_src = rdata + src_inc;
00444 
00445                         if (dst_inc <= (src_inc + civ.x_iter ))
00446                         {
00447                                 // Overlap
00448                                 if ( dst_inc > src_inc )
00449                                 {
00450                                         // Because the memcpy operation is the forward direction, and this "reverse
00451                                         // direction" loop is proceeding in a backwards direction, it is possible
00452                                         // that memory copied to is yet to be copied from (because memcpy goes forward).
00453                                         // In this scenario pixel memory "copied to" is yet to be "copied from"
00454                                         // i.e. there is overlap
00455 
00456                                         // memmove handles overlapping cases.
00457                                         // memmove could use a temporary buffer, or could go just go backwards
00458                                         // the specification doesn't say how the function behaves...
00459                                         // If memmove creates a temporary buffer is clip_inplace no longer inplace?
00460                                         memmove(local_dst, local_src, clipped_row_size);
00461                                 }
00462                                 continue;
00463                         }
00464 
00465                         // This situation not encountered in testing thus far
00466                         Assert( dst_inc < new_size && src_inc < prev_size && dst_inc >= 0 && src_inc >= 0 );
00467 
00468                         // Perform the memory copy
00469                         EMUtil::em_memcpy(local_dst, local_src, clipped_row_size);
00470                 }
00471         }
00472 
00473         // Resize the volume after memory shifting occurs if the new volume is smaller than the previous volume
00474         // set_size calls realloc, guaranteeing that the pixel rdata is in the right location.
00475         if ( new_size < prev_size )
00476                 set_size(new_nx, new_ny, new_nz);
00477 
00478         // Now set all the edges to zero
00479 
00480         // Set the extra bottom z slices to the fill_value
00481         if (  z0 < 0 )
00482         {
00483                 //EMUtil::em_memset(rdata, 0, (-z0)*new_sec_size*sizeof(float));
00484                 size_t inc = (-z0)*new_sec_size;
00485                 std::fill(rdata,rdata+inc,fill_value);
00486         }
00487 
00488         // Set the extra top z slices to the fill_value
00489         if (  civ.new_z_top > 0 )
00490         {
00491                 float* begin_pointer = rdata + (new_nz-civ.new_z_top)*new_sec_size;
00492                 //EMUtil::em_memset(begin_pointer, 0, (civ.new_z_top)*new_sec_size*sizeof(float));
00493                 float* end_pointer = begin_pointer+(civ.new_z_top)*new_sec_size;
00494                 std::fill(begin_pointer,end_pointer,fill_value);
00495         }
00496 
00497         // Next deal with x and y edges by iterating through each slice
00498         for ( int i = civ.new_z_bottom; i < civ.new_z_bottom + civ.z_iter; ++i )
00499         {
00500                 // Set the extra front y components to the fill_value
00501                 if ( y0 < 0 )
00502                 {
00503                         float* begin_pointer = rdata + i*new_sec_size;
00504                         //EMUtil::em_memset(begin_pointer, 0, (-y0)*new_nx*sizeof(float));
00505                         float* end_pointer = begin_pointer+(-y0)*new_nx;
00506                         std::fill(begin_pointer,end_pointer,fill_value);
00507                 }
00508 
00509                 // Set the extra back y components to the fill_value
00510                 if ( civ.new_y_back > 0 )
00511                 {
00512                         float* begin_pointer = rdata + i*new_sec_size + (new_ny-civ.new_y_back)*new_nx;
00513                         //EMUtil::em_memset(begin_pointer, 0, (civ.new_y_back)*new_nx*sizeof(float));
00514                         float* end_pointer = begin_pointer+(civ.new_y_back)*new_nx;
00515                         std::fill(begin_pointer,end_pointer,fill_value);
00516                 }
00517 
00518                 // Iterate through the y to set each correct x component to the fill_value
00519                 for (int j = civ.new_y_front; j <civ.new_y_front + civ.y_iter; ++j)
00520                 {
00521                         // Set the extra left x components to the fill_value
00522                         if ( x0 < 0 )
00523                         {
00524                                 float* begin_pointer = rdata + i*new_sec_size + j*new_nx;
00525                                 //EMUtil::em_memset(begin_pointer, 0, (-x0)*sizeof(float));
00526                                 float* end_pointer = begin_pointer+(-x0);
00527                                 std::fill(begin_pointer,end_pointer,fill_value);
00528                         }
00529 
00530                         // Set the extra right x components to the fill_value
00531                         if ( civ.new_x_right > 0 )
00532                         {
00533                                 float* begin_pointer = rdata + i*new_sec_size + j*new_nx + (new_nx - civ.new_x_right);
00534                                 //EMUtil::em_memset(begin_pointer, 0, (civ.new_x_right)*sizeof(float));
00535                                 float* end_pointer = begin_pointer+(civ.new_x_right);
00536                                 std::fill(begin_pointer,end_pointer,fill_value);
00537                         }
00538 
00539                 }
00540         }
00541 
00542 // These couts may be useful
00543 //      cout << "start starts " << civ.prv_x_left << " " << civ.prv_y_front << " " << civ.prv_z_bottom << endl;
00544 //      cout << "start ends " << civ.prv_x_right << " " << civ.prv_y_back << " " << civ.prv_z_top << endl;
00545 //      cout << "dst starts " << civ.new_x_left << " " << civ.new_y_front << " " << civ.new_z_bottom << endl;
00546 //      cout << "dst ends " << civ.new_x_right << " " << civ.new_y_back << " " << civ.new_z_top << endl;
00547 //      cout << "total iter z - " << civ.z_iter << " y - " << civ.y_iter << " x - " << civ.x_iter << endl;
00548 //      cout << "=====" << endl;
00549 //      cout << "dst_end is " << dst_it_end << " src end is " << src_it_end << endl;
00550 //      cout << "dst_begin is " << dst_it_begin << " src begin is " << src_it_begin << endl;
00551 
00552         // Update appropriate attributes (Copied and pasted from get_clip)
00553         if( attr_dict.has_key("origin_x") && attr_dict.has_key("origin_y") &&
00554         attr_dict.has_key("origin_z") )
00555         {
00556                 float xorigin = attr_dict["origin_x"];
00557                 float yorigin = attr_dict["origin_y"];
00558                 float zorigin = attr_dict["origin_z"];
00559 
00560                 float apix_x = attr_dict["apix_x"];
00561                 float apix_y = attr_dict["apix_y"];
00562                 float apix_z = attr_dict["apix_z"];
00563 
00564                 set_xyz_origin(xorigin + apix_x * area.origin[0],
00565                         yorigin + apix_y * area.origin[1],
00566                         zorigin + apix_z * area.origin[2]);
00567         }
00568 
00569         // Set the update flag because the size of the image has changed and stats should probably be recalculated if requested.
00570         update();
00571 
00572         EXITFUNC;
00573 }
00574 
00575 EMData *EMData::get_clip(const Region & area, const float fill) const
00576 {
00577         ENTERFUNC;
00578         if (get_ndim() != area.get_ndim()) {
00579                 LOGERR("cannot get %dD clip out of %dD image", area.get_ndim(),get_ndim());
00580                 return 0;
00581         }
00582 
00583         EMData *result = new EMData();
00584 
00585         // Ensure that all of the metadata of this is stored in the new object
00586         // Originally added to ensure that euler angles were retained when preprocessing (zero padding) images
00587         // prior to insertion into the 3D for volume in the reconstruction phase (see reconstructor.cpp/h).
00588         result->attr_dict = this->attr_dict;
00589         int zsize = (int)area.size[2];
00590         if (zsize == 0 && nz <= 1) {
00591                 zsize = 1;
00592         }
00593         int ysize = (ny<=1 && (int)area.size[1]==0 ? 1 : (int)area.size[1]);
00594 
00595         if ( (int)area.size[0] < 0 || ysize < 0 || zsize < 0 )
00596         {
00597                 // Negative image dimensions not supported - added retrospectively by d.woolford (who didn't write get_clip but wrote clip_inplace)
00598                 throw ImageDimensionException("New image dimensions are negative - this is not supported in the the get_clip operation");
00599         }
00600 
00601 //#ifdef EMAN2_USING_CUDA
00602         // Strategy is always to prefer using the GPU if possible
00603 //      bool use_gpu = false;
00604 //      if ( gpu_operation_preferred() ) {
00605 //              result->set_size_cuda((int)area.size[0], ysize, zsize);
00606                 //CudaDataLock lock(this); // Just so we never have to recopy this data to and from the GPU
00607 //              result->get_cuda_data(); // Force the allocation - set_size_cuda is lazy
00608                 // Setting the value is necessary seeing as cuda data is not automatically zeroed
00609 //              result->to_value(fill); // This will automatically use the GPU.
00610 //              use_gpu = true;
00611 //      } else { // cpu == True
00612 //              result->set_size((int)area.size[0], ysize, zsize);
00613 //              if (fill != 0.0) { result->to_value(fill); };
00614 //      }
00615 //#else
00616         result->set_size((int)area.size[0], ysize, zsize);
00617         if (fill != 0.0) { result->to_value(fill); };
00618 //#endif //EMAN2_USING_CUDA
00619 
00620         int x0 = (int) area.origin[0];
00621         x0 = x0 < 0 ? 0 : x0;
00622 
00623         int y0 = (int) area.origin[1];
00624         y0 = y0 < 0 ? 0 : y0;
00625 
00626         int z0 = (int) area.origin[2];
00627         z0 = z0 < 0 ? 0 : z0;
00628 
00629         int x1 = (int) (area.origin[0] + area.size[0]);
00630         x1 = x1 > nx ? nx : x1;
00631 
00632         int y1 = (int) (area.origin[1] + area.size[1]);
00633         y1 = y1 > ny ? ny : y1;
00634 
00635         int z1 = (int) (area.origin[2] + area.size[2]);
00636         z1 = z1 > nz ? nz : z1;
00637         if (z1 <= 0) {
00638                 z1 = 1;
00639         }
00640 
00641         result->insert_clip(this,-((IntPoint)area.origin));
00642 
00643         if( attr_dict.has_key("apix_x") && attr_dict.has_key("apix_y") &&
00644                 attr_dict.has_key("apix_z") )
00645         {
00646                 if( attr_dict.has_key("origin_x") && attr_dict.has_key("origin_y") &&
00647                     attr_dict.has_key("origin_z") )
00648                 {
00649                         float xorigin = attr_dict["origin_x"];
00650                         float yorigin = attr_dict["origin_y"];
00651                         float zorigin = attr_dict["origin_z"];
00652 
00653                         float apix_x = attr_dict["apix_x"];
00654                         float apix_y = attr_dict["apix_y"];
00655                         float apix_z = attr_dict["apix_z"];
00656 
00657                         result->set_xyz_origin(xorigin + apix_x * area.origin[0],
00658                                                                    yorigin + apix_y * area.origin[1],
00659                                                                zorigin + apix_z * area.origin[2]);
00660                 }
00661         }
00662 
00663 //#ifdef EMAN2_USING_CUDA
00664 //      if (use_gpu) result->gpu_update();
00665 //      else result->update();
00666 //#else
00667         result->update();
00668 //#endif // EMAN2_USING_CUDA
00669 
00670 
00671         result->set_path(path);
00672         result->set_pathnum(pathnum);
00673 
00674         EXITFUNC;
00675         return result;
00676 }
00677 
00678 
00679 EMData *EMData::get_top_half() const
00680 {
00681         ENTERFUNC;
00682 
00683         if (get_ndim() != 3) {
00684                 throw ImageDimensionException("3D only");
00685         }
00686 
00687         EMData *half = new EMData();
00688         half->attr_dict = attr_dict;
00689         half->set_size(nx, ny, nz / 2);
00690 
00691         float *half_data = half->get_data();
00692         EMUtil::em_memcpy(half_data, &(get_data()[(size_t)nz / 2 * (size_t)nx * (size_t)ny]), sizeof(float) * (size_t)nx * (size_t)ny * (size_t)nz / 2lu);
00693 
00694         float apix_z = attr_dict["apix_z"];
00695         float origin_z = attr_dict["origin_z"];
00696         origin_z += apix_z * nz / 2;
00697         half->attr_dict["origin_z"] = origin_z;
00698         half->update();
00699 
00700         EXITFUNC;
00701         return half;
00702 }
00703 
00704 
00705 EMData *EMData::get_rotated_clip(const Transform &xform,
00706                                                                  const IntSize &size, float)
00707 {
00708         EMData *result = new EMData();
00709         result->set_size(size[0],size[1],size[2]);
00710 
00711         if (nz==1) {
00712                 for (int y=-size[1]/2; y<(size[1]+1)/2; y++) {
00713                         for (int x=-size[0]/2; x<(size[0]+1)/2; x++) {
00714                                 Vec3f xv=xform.transform(Vec3f((float)x,(float)y,0.0f));
00715                                 float v = 0;
00716 
00717                                 if (xv[0]<0||xv[1]<0||xv[0]>nx-2||xv[1]>ny-2) v=0.;
00718                                 else v=sget_value_at_interp(xv[0],xv[1]);
00719                                 result->set_value_at(x+size[0]/2,y+size[1]/2,v);
00720                         }
00721                 }
00722         }
00723         else {
00724                 for (int z=-size[2]/2; z<(size[2]+1)/2; z++) {
00725                         for (int y=-size[1]/2; y<(size[1]+1)/2; y++) {
00726                                 for (int x=-size[0]/2; x<(size[0]+1)/2; x++) {
00727                                         Vec3f xv=xform.transform(Vec3f((float)x,(float)y,0.0f));
00728                                         float v = 0;
00729 
00730                                         if (xv[0]<0||xv[1]<0||xv[2]<0||xv[0]>nx-2||xv[1]>ny-2||xv[2]>nz-2) v=0.;
00731                                         else v=sget_value_at_interp(xv[0],xv[1],xv[2]);
00732                                         result->set_value_at(x+size[0]/2,y+size[1]/2,z+size[2]/2,v);
00733                                 }
00734                         }
00735                 }
00736         }
00737         result->update();
00738 
00739         return result;
00740 }
00741 
00742 
00743 EMData* EMData::window_center(int l) {
00744         ENTERFUNC;
00745         // sanity checks
00746         int n = nx;
00747         if (is_complex()) {
00748                 LOGERR("Need real-space data for window_center()");
00749                 throw ImageFormatException(
00750                         "Complex input image; real-space expected.");
00751         }
00752         if (is_fftpadded()) {
00753                 // image has been fft-padded, compute the real-space size
00754                 n -= (2 - int(is_fftodd()));
00755         }
00756         int corner = n/2 - l/2;
00757         int ndim = get_ndim();
00758         EMData* ret;
00759         switch (ndim) {
00760                 case 3:
00761                         if ((n != ny) || (n != nz)) {
00762                                 LOGERR("Need the real-space image to be cubic.");
00763                                 throw ImageFormatException(
00764                                                 "Need cubic real-space image.");
00765                         }
00766                         ret = get_clip(Region(corner, corner, corner, l, l, l));
00767                         break;
00768                 case 2:
00769                         if (n != ny) {
00770                                 LOGERR("Need the real-space image to be square.");
00771                                 throw ImageFormatException(
00772                                                 "Need square real-space image.");
00773                         }
00774                         //cout << "Using corner " << corner << endl;
00775                         ret = get_clip(Region(corner, corner, l, l));
00776                         break;
00777                 case 1:
00778                         ret = get_clip(Region(corner, l));
00779                         break;
00780                 default:
00781                         throw ImageDimensionException(
00782                                         "window_center only supports 1-d, 2-d, and 3-d images");
00783         }
00784         return ret;
00785         EXITFUNC;
00786 }
00787 
00788 
00789 float *EMData::setup4slice(bool redo)
00790 {
00791         ENTERFUNC;
00792 
00793         if (!is_complex()) {
00794                 throw ImageFormatException("complex image only");
00795         }
00796 
00797         if (get_ndim() != 3) {
00798                 throw ImageDimensionException("3D only");
00799         }
00800 
00801         if (supp) {
00802                 if (redo) {
00803                         EMUtil::em_free(supp);
00804                         supp = 0;
00805                 }
00806                 else {
00807                         EXITFUNC;
00808                         return supp;
00809                 }
00810         }
00811 
00812         const int SUPP_ROW_SIZE = 8;
00813         const int SUPP_ROW_OFFSET = 4;
00814         const int supp_size = SUPP_ROW_SIZE + SUPP_ROW_OFFSET;
00815 
00816         supp = (float *) EMUtil::em_calloc(supp_size * ny * nz, sizeof(float));
00817         int nxy = nx * ny;
00818         int supp_xy = supp_size * ny;
00819         float * data = get_data();
00820 
00821         for (int z = 0; z < nz; z++) {
00822                 size_t cur_z1 = z * nxy;
00823                 size_t cur_z2 = z * supp_xy;
00824 
00825                 for (int y = 0; y < ny; y++) {
00826                         size_t cur_y1 = y * nx;
00827                         size_t cur_y2 = y * supp_size;
00828 
00829                         for (int x = 0; x < SUPP_ROW_SIZE; x++) {
00830                                 size_t k = (x + SUPP_ROW_OFFSET) + cur_y2 + cur_z2;
00831                                 supp[k] = data[x + cur_y1 + cur_z1];
00832                         }
00833                 }
00834         }
00835 
00836         for (int z = 1, zz = nz - 1; z < nz; z++, zz--) {
00837                 size_t cur_z1 = zz * nxy;
00838                 size_t cur_z2 = z * supp_xy;
00839 
00840                 for (int y = 1, yy = ny - 1; y < ny; y++, yy--) {
00841                         supp[y * 12 + cur_z2] = data[4 + yy * nx + cur_z1];
00842                         supp[1 + y * 12 + cur_z2] = -data[5 + yy * nx + cur_z1];
00843                         supp[2 + y * 12 + cur_z2] = data[2 + yy * nx + cur_z1];
00844                         supp[3 + y * 12 + cur_z2] = -data[3 + yy * nx + cur_z1];
00845                 }
00846         }
00847 
00848         EXITFUNC;
00849         return supp;
00850 }
00851 
00852 
00853 void EMData::scale(float s)
00854 {
00855         ENTERFUNC;
00856         Transform t;
00857         t.set_scale(s);
00858         transform(t);
00859         EXITFUNC;
00860 }
00861 
00862 
00863 void EMData::translate(int dx, int dy, int dz)
00864 {
00865         ENTERFUNC;
00866         translate(Vec3i(dx, dy, dz));
00867         EXITFUNC;
00868 }
00869 
00870 
00871 void EMData::translate(float dx, float dy, float dz)
00872 {
00873         ENTERFUNC;
00874         int dx_ = Util::round(dx);
00875         int dy_ = Util::round(dy);
00876         int dz_ = Util::round(dz);
00877         if( ( (dx-dx_) == 0 ) && ( (dy-dy_) == 0 ) && ( (dz-dz_) == 0 )) {
00878                 translate(dx_, dy_, dz_);
00879         }
00880         else {
00881                 translate(Vec3f(dx, dy, dz));
00882         }
00883         EXITFUNC;
00884 }
00885 
00886 
00887 void EMData::translate(const Vec3i &translation)
00888 {
00889         ENTERFUNC;
00890 
00891         //if traslation is 0, do nothing
00892         if( translation[0] == 0 && translation[1] == 0 && translation[2] == 0) {
00893                 EXITFUNC;
00894                 return;
00895         }
00896 
00897         Dict params("trans",static_cast< vector<int> >(translation));
00898         process_inplace("math.translate.int",params);
00899 
00900         // update() - clip_inplace does the update
00901         all_translation += translation;
00902 
00903         EXITFUNC;
00904 }
00905 
00906 
00907 void EMData::translate(const Vec3f &translation)
00908 {
00909         ENTERFUNC;
00910 
00911         if( translation[0] == 0.0f && translation[1] == 0.0f && translation[2] == 0.0f ) {
00912                 EXITFUNC;
00913                 return;
00914         }
00915 
00916         Transform* t = new Transform();
00917         t->set_trans(translation);
00918         process_inplace("xform",Dict("transform",t));
00919         delete t;
00920 
00921         all_translation += translation;
00922         EXITFUNC;
00923 }
00924 
00925 
00926 void EMData::rotate(float az, float alt, float phi)
00927 {
00928         Dict d("type","eman");
00929         d["az"] = az;
00930         d["alt"] = alt;
00931         d["phi"] = phi;
00932         Transform t(d);
00933         transform(t);
00934 }
00935 
00936 
00937 
00938 void EMData::rotate(const Transform & t)
00939 {
00940         cout << "Deprecation warning in EMData::rotate. Please consider using EMData::transform() instead " << endl;
00941         transform(t);
00942 }
00943 
00944 float EMData::max_3D_pixel_error(const Transform &t1, const Transform & t2, float r) {
00945         
00946         Transform t;
00947         int r0 = (int)r;
00948         float ddmax = 0.0f;
00949 
00950         t = t2*t1.inverse();
00951         for (int i=0; i<int(2*M_PI*r0+0.5); i++) {
00952                 Vec3f v = Vec3f(r0*cos((float)i), r0*sin((float)i), 0);
00953                 Vec3f d = t*v-v;
00954                 float dd = d[0]*d[0]+d[1]*d[1]+d[2]*d[2];
00955                 if (dd > ddmax) ddmax = dd; 
00956         }
00957         return std::sqrt(ddmax);
00958 }
00959 
00960 void EMData::rotate_translate(float az, float alt, float phi, float dx, float dy, float dz)
00961 {
00962         cout << "Deprecation warning in EMData::rotate_translate. Please consider using EMData::transform() instead " << endl;
00963 //      Transform3D t( az, alt, phi,Vec3f(dx, dy, dz));
00964         Transform t;
00965         t.set_rotation(Dict("type", "eman", "az", az, "alt", alt, "phi", phi));
00966         t.set_trans(dx, dy, dz);
00967         rotate_translate(t);
00968 }
00969 
00970 
00971 void EMData::rotate_translate(float az, float alt, float phi, float dx, float dy,
00972                                                           float dz, float pdx, float pdy, float pdz)
00973 {
00974         cout << "Deprecation warning in EMData::rotate_translate. Please consider using EMData::transform() instead " << endl;
00975 //      Transform3D t(Vec3f(dx, dy, dz), az, alt, phi, Vec3f(pdx,pdy,pdz));
00976 //      rotate_translate(t);
00977 
00978         Transform t;
00979         t.set_pre_trans(Vec3f(dx, dy, dz));
00980         t.set_rotation(Dict("type", "eman", "az", az, "alt", alt, "phi", phi));
00981         t.set_trans(pdx, pdy, pdz);
00982         rotate_translate(t);
00983 }
00984 
00985 //void EMData::rotate_translate(const Transform3D & RA)
00986 //{
00987 //      cout << "Deprecation warning in EMData::rotate_translate. Please consider using EMData::transform() instead " << endl;
00988 //      ENTERFUNC;
00989 //
00990 //#if EMDATA_EMAN2_DEBUG
00991 //      std::cout << "start rotate_translate..." << std::endl;
00992 //#endif
00993 //
00994 //      float scale       = RA.get_scale();
00995 //      Vec3f dcenter     = RA.get_center();
00996 //      Vec3f translation = RA.get_posttrans();
00997 //      Dict rotation      = RA.get_rotation(Transform3D::EMAN);
00999 //      Transform3D RAInv = RA.inverse(); // We're rotating the coordinate system, not the data
01001 //#if EMDATA_EMAN2_DEBUG
01002 //      vector<string> keys = rotation.keys();
01003 //      vector<string>::const_iterator it;
01004 //      for(it=keys.begin(); it!=keys.end(); ++it) {
01006 //              std::cout << *it << " : " << (float)rotation.get(*it) << std::endl;
01007 //      }
01008 //#endif
01009 //      float inv_scale = 1.0f;
01010 //
01011 //      if (scale != 0) {
01012 //              inv_scale = 1.0f / scale;
01013 //      }
01014 //
01015 //      float *src_data = 0;
01016 //      float *des_data = 0;
01017 //
01018 //      src_data = get_data();
01019 //      des_data = (float *) EMUtil::em_malloc(nx * ny * nz * sizeof(float));
01020 //
01021 //      if (nz == 1) {
01022 //              float x2c =  nx / 2 - dcenter[0] + RAInv[0][3];
01023 //              float y2c =  ny / 2 - dcenter[1] + RAInv[1][3];
01024 //              float y   = -ny / 2 + dcenter[1]; // changed 0 to 1 in dcenter and below
01025 //              for (int j = 0; j < ny; j++, y += 1.0f) {
01026 //                      float x = -nx / 2 + dcenter[0];
01027 //                      for (int i = 0; i < nx; i++, x += 1.0f) {
01028 //                              float x2 = RAInv[0][0]*x + RAInv[0][1]*y + x2c;
01029 //                              float y2 = RAInv[1][0]*x + RAInv[1][1]*y + y2c;
01030 //
01031 //                              if (x2 < 0 || x2 >= nx || y2 < 0 || y2 >= ny ) {
01032 //                                      des_data[i + j * nx] = 0; // It may be tempting to set this value to the
01033 //                                      // mean but in fact this is not a good thing to do. Talk to S.Ludtke about it.
01034 //                              }
01035 //                              else {
01036 //                                      int ii = Util::fast_floor(x2);
01037 //                                      int jj = Util::fast_floor(y2);
01038 //                                      int k0 = ii + jj * nx;
01039 //                                      int k1 = k0 + 1;
01040 //                                      int k2 = k0 + nx;
01041 //                                      int k3 = k0 + nx + 1;
01042 //
01043 //                                      if (ii == nx - 1) {
01044 //                                              k1--;
01045 //                                              k3--;
01046 //                                      }
01047 //                                      if (jj == ny - 1) {
01048 //                                              k2 -= nx;
01049 //                                              k3 -= nx;
01050 //                                      }
01051 //
01052 //                                      float t = x2 - ii;
01053 //                                      float u = y2 - jj;
01054 //
01055 //                                      des_data[i + j * nx] = Util::bilinear_interpolate(src_data[k0],src_data[k1], src_data[k2], src_data[k3],t,u); // This is essentially basic interpolation
01056 //                              }
01057 //                      }
01058 //              }
01059 //      }
01060 //      else {
01061 //#if EMDATA_EMAN2_DEBUG
01062 //              std::cout << "This is the 3D case." << std::endl    ;
01063 //#endif
01064 //
01065 //              Transform3D mx = RA;
01066 //              mx.set_scale(inv_scale);
01067 //
01068 //#if EMDATA_EMAN2_DEBUG
01071 //#endif
01072 //
01073 //              int nxy = nx * ny;
01074 //              int l = 0;
01075 //
01076 //              float x2c =  nx / 2 - dcenter[0] + RAInv[0][3];;
01077 //              float y2c =  ny / 2 - dcenter[1] + RAInv[1][3];;
01078 //              float z2c =  nz / 2 - dcenter[2] + RAInv[2][3];;
01079 //
01080 //              float z   = -nz / 2 + dcenter[2]; //
01081 //
01082 //              size_t ii, k0, k1, k2, k3, k4, k5, k6, k7;
01083 //              for (int k = 0; k < nz; k++, z += 1.0f) {
01084 //                      float y   = -ny / 2 + dcenter[1]; //
01085 //                      for (int j = 0; j < ny; j++,   y += 1.0f) {
01086 //                              float x = -nx / 2 + dcenter[0];
01087 //                              for (int i = 0; i < nx; i++, l++ ,  x += 1.0f) {
01088 //                                      float x2 = RAInv[0][0] * x + RAInv[0][1] * y + RAInv[0][2] * z + x2c;
01089 //                                      float y2 = RAInv[1][0] * x + RAInv[1][1] * y + RAInv[1][2] * z + y2c;
01090 //                                      float z2 = RAInv[2][0] * x + RAInv[2][1] * y + RAInv[2][2] * z + z2c;
01091 //
01092 //
01093 //                                      if (x2 < 0 || y2 < 0 || z2 < 0 ||
01094 //                                              x2 >= nx  || y2 >= ny  || z2>= nz ) {
01095 //                                              des_data[l] = 0;
01096 //                                      }
01097 //                                      else {
01098 //                                              int ix = Util::fast_floor(x2);
01099 //                                              int iy = Util::fast_floor(y2);
01100 //                                              int iz = Util::fast_floor(z2);
01101 //                                              float tuvx = x2-ix;
01102 //                                              float tuvy = y2-iy;
01103 //                                              float tuvz = z2-iz;
01104 //                                              ii = ix + iy * nx + iz * nxy;
01105 //
01106 //                                              k0 = ii;
01107 //                                              k1 = k0 + 1;
01108 //                                              k2 = k0 + nx;
01109 //                                              k3 = k0 + nx+1;
01110 //                                              k4 = k0 + nxy;
01111 //                                              k5 = k1 + nxy;
01112 //                                              k6 = k2 + nxy;
01113 //                                              k7 = k3 + nxy;
01114 //
01115 //                                              if (ix == nx - 1) {
01116 //                                                      k1--;
01117 //                                                      k3--;
01118 //                                                      k5--;
01119 //                                                      k7--;
01120 //                                              }
01121 //                                              if (iy == ny - 1) {
01122 //                                                      k2 -= nx;
01123 //                                                      k3 -= nx;
01124 //                                                      k6 -= nx;
01125 //                                                      k7 -= nx;
01126 //                                              }
01127 //                                              if (iz == nz - 1) {
01128 //                                                      k4 -= nxy;
01129 //                                                      k5 -= nxy;
01130 //                                                      k6 -= nxy;
01131 //                                                      k7 -= nxy;
01132 //                                              }
01133 //
01134 //                                              des_data[l] = Util::trilinear_interpolate(src_data[k0],
01135 //                                                        src_data[k1],
01136 //                                                        src_data[k2],
01137 //                                                        src_data[k3],
01138 //                                                        src_data[k4],
01139 //                                                        src_data[k5],
01140 //                                                        src_data[k6],
01141 //                                                        src_data[k7],
01142 //                                                        tuvx, tuvy, tuvz);
01143 //#if EMDATA_EMAN2_DEBUG
01144 //                                              printf(" ix=%d \t iy=%d \t iz=%d \t value=%f \n", ix ,iy, iz, des_data[l] );
01145 //                                              std::cout << src_data[ii] << std::endl;
01146 //#endif
01147 //                                      }
01148 //                              }
01149 //                      }
01150 //              }
01151 //      }
01152 //
01153 //      if( rdata )
01154 //      {
01155 //              EMUtil::em_free(rdata);
01156 //              rdata = 0;
01157 //      }
01158 //      rdata = des_data;
01159 //
01160 //      scale_pixel(inv_scale);
01161 //
01162 //      attr_dict["origin_x"] = (float) attr_dict["origin_x"] * inv_scale;
01163 //      attr_dict["origin_y"] = (float) attr_dict["origin_y"] * inv_scale;
01164 //      attr_dict["origin_z"] = (float) attr_dict["origin_z"] * inv_scale;
01165 //
01166 //      update();
01167 //      all_translation += translation;
01168 //      EXITFUNC;
01169 //}
01170 
01171 
01172 
01173 
01174 void EMData::rotate_x(int dx)
01175 {
01176         ENTERFUNC;
01177 
01178         if (get_ndim() > 2) {
01179                 throw ImageDimensionException("no 3D image");
01180         }
01181 
01182 
01183         size_t row_size = nx * sizeof(float);
01184         float *tmp = (float*)EMUtil::em_malloc(row_size);
01185         float * data = get_data();
01186 
01187         for (int y = 0; y < ny; y++) {
01188                 int y_nx = y * nx;
01189                 for (int x = 0; x < nx; x++) {
01190                         tmp[x] = data[y_nx + (x + dx) % nx];
01191                 }
01192                 EMUtil::em_memcpy(&data[y_nx], tmp, row_size);
01193         }
01194 
01195         update();
01196         if( tmp )
01197         {
01198                 delete[]tmp;
01199                 tmp = 0;
01200         }
01201         EXITFUNC;
01202 }
01203 
01204 double EMData::dot_rotate_translate(EMData * with, float dx, float dy, float da, const bool mirror)
01205 {
01206         ENTERFUNC;
01207 
01208         if (!EMUtil::is_same_size(this, with)) {
01209                 LOGERR("images not same size");
01210                 throw ImageFormatException("images not same size");
01211         }
01212 
01213         if (get_ndim() == 3) {
01214                 LOGERR("1D/2D Images only");
01215                 throw ImageDimensionException("1D/2D only");
01216         }
01217 
01218         float *this_data = 0;
01219 
01220         this_data = get_data();
01221 
01222         float da_rad = da*(float)M_PI/180.0f;
01223 
01224         float *with_data = with->get_data();
01225         float mx0 = cos(da_rad);
01226         float mx1 = sin(da_rad);
01227         float y = -ny / 2.0f;
01228         float my0 = mx0 * (-nx / 2.0f - 1.0f) + nx / 2.0f - dx;
01229         float my1 = -mx1 * (-nx / 2.0f - 1.0f) + ny / 2.0f - dy;
01230         double result = 0;
01231 
01232         for (int j = 0; j < ny; j++) {
01233                 float x2 = my0 + mx1 * y;
01234                 float y2 = my1 + mx0 * y;
01235 
01236                 int ii = Util::fast_floor(x2);
01237                 int jj = Util::fast_floor(y2);
01238                 float t = x2 - ii;
01239                 float u = y2 - jj;
01240 
01241                 for (int i = 0; i < nx; i++) {
01242                         t += mx0;
01243                         u -= mx1;
01244 
01245                         if (t >= 1.0f) {
01246                                 ii++;
01247                                 t -= 1.0f;
01248                         }
01249 
01250                         if (u >= 1.0f) {
01251                                 jj++;
01252                                 u -= 1.0f;
01253                         }
01254 
01255                         if (t < 0) {
01256                                 ii--;
01257                                 t += 1.0f;
01258                         }
01259 
01260                         if (u < 0) {
01261                                 jj--;
01262                                 u += 1.0f;
01263                         }
01264 
01265                         if (ii >= 0 && ii <= nx - 2 && jj >= 0 && jj <= ny - 2) {
01266                                 int k0 = ii + jj * nx;
01267                                 int k1 = k0 + 1;
01268                                 int k2 = k0 + nx + 1;
01269                                 int k3 = k0 + nx;
01270 
01271                                 float tt = 1 - t;
01272                                 float uu = 1 - u;
01273                                 int idx = i + j * nx;
01274                                 if (mirror) idx = nx-1-i+j*nx; // mirroring of Transforms is always about the y axis
01275                                 result += (this_data[k0] * tt * uu + this_data[k1] * t * uu +
01276                                                    this_data[k2] * t * u + this_data[k3] * tt * u) * with_data[idx];
01277                         }
01278                 }
01279                 y += 1.0f;
01280         }
01281 
01282         EXITFUNC;
01283         return result;
01284 }
01285 
01286 
01287 EMData *EMData::little_big_dot(EMData * with, bool do_sigma)
01288 {
01289         ENTERFUNC;
01290 
01291         if (get_ndim() > 2) {
01292                 throw ImageDimensionException("1D/2D only");
01293         }
01294 
01295         EMData *ret = copy_head();
01296         ret->set_size(nx,ny,nz);
01297         ret->to_zero();
01298 
01299         int nx2 = with->get_xsize();
01300         int ny2 = with->get_ysize();
01301         float em = with->get_edge_mean();
01302 
01303         float *data = get_data();
01304         float *with_data = with->get_data();
01305         float *ret_data = ret->get_data();
01306 
01307         float sum2 = (Util::square((float)with->get_attr("sigma")) +
01308                                   Util::square((float)with->get_attr("mean")));
01309         if (do_sigma) {
01310                 for (int j = ny2 / 2; j < ny - ny2 / 2; j++) {
01311                         for (int i = nx2 / 2; i < nx - nx2 / 2; i++) {
01312                                 float sum = 0;
01313                                 float sum1 = 0;
01314                                 float summ = 0;
01315                                 int k = 0;
01316 
01317                                 for (int jj = j - ny2 / 2; jj < j + ny2 / 2; jj++) {
01318                                         for (int ii = i - nx2 / 2; ii < i + nx2 / 2; ii++) {
01319                                                 int l = ii + jj * nx;
01320                                                 sum1 += Util::square(data[l]);
01321                                                 summ += data[l];
01322                                                 sum += data[l] * with_data[k];
01323                                                 k++;
01324                                         }
01325                                 }
01326                                 float tmp_f1 = (sum1 / 2.0f - sum) / (nx2 * ny2);
01327                                 float tmp_f2 = Util::square((float)with->get_attr("mean") -
01328                                                                                         summ / (nx2 * ny2));
01329                                 ret_data[i + j * nx] = sum2 + tmp_f1 - tmp_f2;
01330                         }
01331                 }
01332         }
01333         else {
01334                 for (int j = ny2 / 2; j < ny - ny2 / 2; j++) {
01335                         for (int i = nx2 / 2; i < nx - nx2 / 2; i++) {
01336                                 float eml = 0;
01337                                 float dot = 0;
01338                                 float dot2 = 0;
01339 
01340                                 for (int ii = i - nx2 / 2; ii < i + nx2 / 2; ii++) {
01341                                         eml += data[ii + (j - ny2 / 2) * nx] + data[ii + (j + ny2 / 2 - 1) * nx];
01342                                 }
01343 
01344                                 for (int jj = j - ny2 / 2; jj < j + ny2 / 2; jj++) {
01345                                         eml += data[i - nx2 / 2 + jj * nx] + data[i + nx2 / 2 - 1 + jj * nx];
01346                                 }
01347 
01348                                 eml /= (nx2 + ny2) * 2.0f;
01349                                 int k = 0;
01350 
01351                                 for (int jj = j - ny2 / 2; jj < j + ny2 / 2; jj++) {
01352                                         for (int ii = i - nx2 / 2; ii < i + nx2 / 2; ii++) {
01353                                                 dot += (data[ii + jj * nx] - eml) * (with_data[k] - em);
01354                                                 dot2 += Util::square(data[ii + jj * nx] - eml);
01355                                                 k++;
01356                                         }
01357                                 }
01358 
01359                                 dot2 = std::sqrt(dot2);
01360 
01361                                 if (dot2 == 0) {
01362                                         ret_data[i + j * nx] = 0;
01363                                 }
01364                                 else {
01365                                         ret_data[i + j * nx] = dot / (nx2 * ny2 * dot2 * (float)with->get_attr("sigma"));
01366                                 }
01367                         }
01368                 }
01369         }
01370 
01371         ret->update();
01372 
01373         EXITFUNC;
01374         return ret;
01375 }
01376 
01377 
01378 EMData *EMData::do_radon()
01379 {
01380         ENTERFUNC;
01381 
01382         if (get_ndim() != 2) {
01383                 throw ImageDimensionException("2D only");
01384         }
01385 
01386         if (nx != ny) {
01387                 throw ImageFormatException("square image only");
01388         }
01389 
01390         EMData *result = new EMData();
01391         result->set_size(nx, ny, 1);
01392         result->to_zero();
01393         float *result_data = result->get_data();
01394 
01395         EMData *this_copy = this;
01396         this_copy = copy();
01397 
01398         for (int i = 0; i < nx; i++) {
01399                 Transform t(Dict("type","2d","alpha",(float) M_PI * 2.0f * i / nx));
01400                 this_copy->transform(t);
01401 
01402                 float *copy_data = this_copy->get_data();
01403 
01404                 for (int y = 0; y < nx; y++) {
01405                         for (int x = 0; x < nx; x++) {
01406                                 if (Util::square(x - nx / 2) + Util::square(y - nx / 2) <= nx * nx / 4) {
01407                                         result_data[i + y * nx] += copy_data[x + y * nx];
01408                                 }
01409                         }
01410                 }
01411 
01412                 this_copy->update();
01413         }
01414 
01415         result->update();
01416 
01417         if( this_copy )
01418         {
01419                 delete this_copy;
01420                 this_copy = 0;
01421         }
01422 
01423         EXITFUNC;
01424         return result;
01425 }
01426 
01427 void EMData::zero_corner_circulant(const int radius)
01428 {
01429         if ( nz > 1 && nz < (2*radius+1) ) throw ImageDimensionException("Error: cannot zero corner - nz is too small");
01430         if ( ny > 1 && ny < (2*radius+1) ) throw ImageDimensionException("Error: cannot zero corner - ny is too small");
01431         if ( nx > 1 && nx < (2*radius+1) ) throw ImageDimensionException("Error: cannot zero corner - nx is too small");
01432 
01433         int it_z = radius;
01434         int it_y = radius;
01435         int it_x = radius;
01436 
01437         if ( nz == 1 ) it_z = 0;
01438         if ( ny == 1 ) it_y = 0;
01439         if ( nx == 1 ) it_z = 0;
01440 
01441         if ( nz == 1 && ny == 1 )
01442         {
01443                 for ( int x = -it_x; x <= it_x; ++x )
01444                         get_value_at_wrap(x) = 0;
01445 
01446         }
01447         else if ( nz == 1 )
01448         {
01449                 for ( int y = -it_y; y <= it_y; ++y)
01450                         for ( int x = -it_x; x <= it_x; ++x )
01451                                 get_value_at_wrap(x,y) = 0;
01452         }
01453         else
01454         {
01455                 for( int z = -it_z; z <= it_z; ++z )
01456                         for ( int y = -it_y; y <= it_y; ++y)
01457                                 for ( int x = -it_x; x < it_x; ++x )
01458                                         get_value_at_wrap(x,y,z) = 0;
01459 
01460         }
01461 
01462 }
01463 
01464 EMData *EMData::calc_ccf(EMData * with, fp_flag fpflag,bool center)
01465 {
01466         ENTERFUNC;
01467         if( with == 0 ) {
01468                 EXITFUNC;
01469                 return convolution(this,this,fpflag, center);
01470         }
01471         else if ( with == this ){ // this if statement is not necessary, the correlation function tests to see if with == this
01472                 EXITFUNC;
01473                 return correlation(this, this, fpflag,center);
01474         }
01475         else {
01476 
01477 #ifdef EMAN2_USING_CUDA //CUDA 
01478                 // assume always get rw data (makes life a lot easier!!! 
01479                 // also assume that both images are the same size. When using CUDA we are only interested in speed, not flexibility!!
01480                 // P.S. (I feel like I am pounding square pegs into a round holes with CUDA)
01481                 if(cudarwdata && with->cudarwdata) {
01482                         //cout << "using CUDA for ccf" << endl;
01483                         EMData* afft = 0;
01484                         EMData* bfft = 0;
01485                         bool delafft = false, delbfft = false;
01486                         int offset = 0;
01487                         
01488                         //do ffts if not alreay done
01489                         if(!is_complex()){
01490                                 afft = do_fft_cuda();
01491                                 delafft = true;
01492                                 offset = 2 - nx%2;
01493                                 //cout << "Do cuda FFT A" << endl;
01494                         }else{
01495                                 afft = this;
01496                         }
01497                         if(!with->is_complex()){
01498                                 bfft = with->do_fft_cuda();
01499                                 //cout << "Do cuda FFT B" << endl;
01500                                 delbfft = true;
01501                         }else{
01502                                 bfft = with;
01503                         }
01504 
01505                         calc_ccf_cuda(afft->cudarwdata,bfft->cudarwdata,nx + offset, ny, nz); //this is the business end, results go in afft
01506                         
01507                         if(delbfft) delete bfft;
01508                         
01509                         EMData * corr = afft->do_ift_cuda();
01510                         if(delafft) delete afft;
01511                         //cor->do_ift_inplace_cuda();//a bit faster, but I'll alos need to rearrnage the mem structure for it to work, BUT this is very SLOW.
01512                         
01513                         return corr;
01514                 }
01515 #endif
01516                 
01517                 // If the argument EMData pointer is not the same size we automatically resize it
01518                 bool undoresize = false;
01519                 int wnx = with->get_xsize(); int wny = with->get_ysize(); int wnz = with->get_zsize(); // obviously is one image is complex and the other real they won't be the same size and we DONT! want to clip JFF
01520                 if (!(is_complex() ^ with->is_complex()) && (wnx != nx || wny != ny || wnz != nz)) {
01521                         cout << "Warning!!! resizing image before CCF calculation" << endl;
01522                         Region r((wnx-nx)/2, (wny-ny)/2, (wnz-nz)/2,nx,ny,nz);
01523                         with->clip_inplace(r);
01524                         undoresize = true;
01525                 }
01526 
01527                 EMData* cor = correlation(this, with, fpflag, center);
01528 
01529                 // If the argument EMData pointer was resized, it is returned to its original dimensions
01530                 if ( undoresize ) {
01531                         Region r((nx-wnx)/2, (ny-wny)/2,(nz-wnz)/2,wnx,wny,wnz);
01532                         with->clip_inplace(r);
01533                 }
01534 
01535                 EXITFUNC;
01536                 return cor;
01537         }
01538 }
01539 
01540 EMData *EMData::calc_ccfx( EMData * const with, int y0, int y1, bool no_sum, bool flip)
01541 {
01542         ENTERFUNC;
01543 
01544         if (!with) {
01545                 LOGERR("NULL 'with' image. ");
01546                 throw NullPointerException("NULL input image");
01547         }
01548 
01549         if (!EMUtil::is_same_size(this, with)) {
01550                 LOGERR("images not same size: (%d,%d,%d) != (%d,%d,%d)",
01551                            nx, ny, nz,
01552                            with->get_xsize(), with->get_ysize(), with->get_zsize());
01553                 throw ImageFormatException("images not same size");
01554         }
01555         if (get_ndim() > 2) {
01556                 LOGERR("2D images only");
01557                 throw ImageDimensionException("2D images only");
01558         }
01559 
01560         if (y1 <= y0) {
01561                 y1 = ny;
01562         }
01563 
01564         if (y0 >= y1) {
01565                 y0 = 0;
01566         }
01567 
01568         if (y0 < 0) {
01569                 y0 = 0;
01570         }
01571 
01572         if (y1 > ny) {
01573                 y1 = ny;
01574         }
01575         if (is_complex_x() || with->is_complex_x() ) throw; // Woops don't support this anymore!
01576 
01577         static int nx_fft = 0;
01578         static int ny_fft = 0;
01579         static EMData f1;
01580         static EMData f2;
01581         static EMData rslt;
01582 
01583         int height = y1-y0;
01584         int width = (nx+2-(nx%2));
01585         if (width != nx_fft || height != ny_fft ) {
01586                 f1.set_size(width,height);
01587                 f2.set_size(width,height);
01588                 rslt.set_size(nx,height);
01589                 nx_fft = width;
01590                 ny_fft = height;
01591         }
01592 
01593 #ifdef EMAN2_USING_CUDA
01594         if (cudarwdata && with->cudarwdata) {
01595                 //cout << "calc_ccfx CUDA" << endl;
01596                 if(!f1.cudarwdata) f1.rw_alloc();
01597                 if(!f2.cudarwdata) f2.rw_alloc();
01598                 if(!rslt.cudarwdata) rslt.rw_alloc();
01599                 cuda_dd_fft_real_to_complex_nd(cudarwdata, f1.cudarwdata, nx, 1, 1, height);
01600                 cuda_dd_fft_real_to_complex_nd(with->cudarwdata, f2.cudarwdata, nx, 1, 1, height);
01601                 calc_ccf_cuda(f1.cudarwdata, f2.cudarwdata, nx, ny, nz);
01602                 cuda_dd_fft_complex_to_real_nd(f1.cudarwdata, rslt.cudarwdata, nx, 1, 1, height);
01603                 if(no_sum){
01604                         EMData* result = new EMData(rslt);
01605                         return result;
01606                 }
01607                 EMData* cf = new EMData(0,0,nx,1,1); //cuda constructor
01608                 cf->runcuda(emdata_column_sum(rslt.cudarwdata, nx, ny));
01609 
01610                 EXITFUNC;
01611                 return cf;
01612         }
01613 #endif
01614 
01615         float *d1 = get_data();
01616         float *d2 = with->get_data();
01617         float *f1d = f1.get_data();
01618         float *f2d = f2.get_data();
01619         for (int j = 0; j < height; j++) {
01620                 EMfft::real_to_complex_1d(d1 + j * nx, f1d+j*width, nx);
01621                 EMfft::real_to_complex_1d(d2 + j * nx, f2d+j*width, nx);
01622         }
01623 
01624         if(flip == false) {
01625                 for (int j = 0; j < height; j++) {
01626                         float *f1a = f1d + j * width;
01627                         float *f2a = f2d + j * width;
01628 
01629                         for (int i = 0; i < width / 2; i++) {
01630                                 float re1 = f1a[2*i];
01631                                 float re2 = f2a[2*i];
01632                                 float im1 = f1a[2*i+1];
01633                                 float im2 = f2a[2*i+1];
01634 
01635                                 f1d[j*width+i*2] = re1 * re2 + im1 * im2;
01636                                 f1d[j*width+i*2+1] = im1 * re2 - re1 * im2;
01637                         }
01638                 }
01639         } else {
01640                 for (int j = 0; j < height; j++) {
01641                         float *f1a = f1d + j * width;
01642                         float *f2a = f2d + j * width;
01643 
01644                         for (int i = 0; i < width / 2; i++) {
01645                                 float re1 = f1a[2*i];
01646                                 float re2 = f2a[2*i];
01647                                 float im1 = f1a[2*i+1];
01648                                 float im2 = f2a[2*i+1];
01649 
01650                                 f1d[j*width+i*2] = re1 * re2 - im1 * im2;
01651                                 f1d[j*width+i*2+1] = im1 * re2 + re1 * im2;
01652                         }
01653                 }
01654         }
01655 
01656         float* rd = rslt.get_data();
01657         for (int j = y0; j < y1; j++) {
01658                 EMfft::complex_to_real_1d(f1d+j*width, rd+j*nx, nx);
01659         }
01660 
01661         if (no_sum) {
01662                 rslt.update(); // This is important in terms of the copy - the returned object won't have the correct flags unless we do this
01663                 EXITFUNC;
01664                 return new EMData(rslt);
01665         } else {
01666                 EMData *cf = new EMData(nx,1,1);
01667                 cf->to_zero();
01668                 float *c = cf->get_data();
01669                 for (int j = 0; j < height; j++) {
01670                         for(int i = 0; i < nx; ++i) {
01671                                 c[i] += rd[i+j*nx];
01672                         }
01673                 }
01674                 cf->update();
01675                 EXITFUNC;
01676                 return cf;
01677         }
01678 }
01679 
01680 EMData *EMData::make_rotational_footprint_cmc( bool unwrap) {
01681         ENTERFUNC;
01682         update_stat();
01683         // Note that rotational_footprint caching saves a large amount of time
01684         // but this is at the expense of memory. Note that a policy is hardcoded here,
01685         // that is that caching is only employed when premasked is false and unwrap
01686         // is true - this is probably going to be what is used in most scenarios
01687         // as advised by Steve Ludtke - In terms of performance this caching doubles the metric
01688         // generated by e2speedtest.
01689         if ( rot_fp != 0 && unwrap == true) {
01690                 return new EMData(*rot_fp);
01691         }
01692 
01693         static EMData obj_filt;
01694         EMData* filt = &obj_filt;
01695         filt->set_complex(true);
01696 
01697 
01698         // The filter object is nothing more than a cached high pass filter
01699         // Ultimately it is used an argument to the EMData::mult(EMData,prevent_complex_multiplication (bool))
01700         // function in calc_mutual_correlation. Note that in the function the prevent_complex_multiplication
01701         // set to true, which is used for speed reasons.
01702         if (filt->get_xsize() != nx+2-(nx%2) || filt->get_ysize() != ny ||
01703                    filt->get_zsize() != nz ) {
01704                 filt->set_size(nx+2-(nx%2), ny, nz);
01705                 filt->to_one();
01706 
01707                 filt->process_inplace("filter.highpass.gauss", Dict("cutoff_abs", 1.5f/nx));
01708         }
01709 
01710         EMData *ccf = this->calc_mutual_correlation(this, true,filt);
01711         ccf->sub(ccf->get_edge_mean());
01712         EMData *result = ccf->unwrap();
01713         delete ccf; ccf = 0;
01714 
01715         EXITFUNC;
01716         if ( unwrap == true)
01717         {
01718         // this if statement reflects a strict policy of caching in only one scenario see comments at beginning of function block
01719 
01720 // Note that the if statement at the beginning of this function ensures that rot_fp is not zero, so there is no need
01721 // to throw any exception
01722 // if ( rot_fp != 0 ) throw UnexpectedBehaviorException("The rotational foot print is only expected to be cached if it is not NULL");
01723 
01724 // Here is where the caching occurs - the rot_fp takes ownsherhip of the pointer, and a deep copied EMData object is returned.
01725 // The deep copy invokes a cost in terms of CPU cycles and memory, but prevents the need for complicated memory management (reference counting)
01726                 rot_fp = result;
01727                 return new EMData(*rot_fp);
01728         }
01729         else return result;
01730 }
01731 
01732 EMData *EMData::make_rotational_footprint( bool unwrap) {
01733         ENTERFUNC;
01734         update_stat();
01735         // Note that rotational_footprint caching saves a large amount of time
01736         // but this is at the expense of memory. Note that a policy is hardcoded here,
01737         // that is that caching is only employed when premasked is false and unwrap
01738         // is true - this is probably going to be what is used in most scenarios
01739         // as advised by Steve Ludtke - In terms of performance this caching doubles the metric
01740         // generated by e2speedtest.
01741         if ( rot_fp != 0 && unwrap == true) {
01742                 return new EMData(*rot_fp);
01743         }
01744 
01745         EMData* ccf = this->calc_ccf(this,CIRCULANT,true);
01746         ccf->sub(ccf->get_edge_mean());
01747         //ccf->process_inplace("xform.phaseorigin.tocenter"); ccf did the centering
01748         EMData *result = ccf->unwrap();
01749         delete ccf; ccf = 0;
01750 
01751         EXITFUNC;
01752         if ( unwrap == true)
01753         { // this if statement reflects a strict policy of caching in only one scenario see comments at beginning of function block
01754 
01755 // Note that the if statement at the beginning of this function ensures that rot_fp is not zero, so there is no need
01756 // to throw any exception
01757 // if ( rot_fp != 0 ) throw UnexpectedBehaviorException("The rotational foot print is only expected to be cached if it is not NULL");
01758 
01759 // Here is where the caching occurs - the rot_fp takes ownsherhip of the pointer, and a deep copied EMData object is returned.
01760 // The deep copy invokes a cost in terms of CPU cycles and memory, but prevents the need for complicated memory management (reference counting)
01761                 rot_fp = result;
01762                 return new EMData(*rot_fp);
01763         }
01764         else return result;
01765 }
01766 
01767 EMData *EMData::make_rotational_footprint_e1( bool unwrap)
01768 {
01769         ENTERFUNC;
01770 
01771         update_stat();
01772         // Note that rotational_footprint caching saves a large amount of time
01773         // but this is at the expense of memory. Note that a policy is hardcoded here,
01774         // that is that caching is only employed when premasked is false and unwrap
01775         // is true - this is probably going to be what is used in most scenarios
01776         // as advised by Steve Ludtke - In terms of performance this caching doubles the metric
01777         // generated by e2speedtest.
01778         if ( rot_fp != 0 && unwrap == true) {
01779                 return new EMData(*rot_fp);
01780         }
01781 
01782         static EMData obj_filt;
01783         EMData* filt = &obj_filt;
01784         filt->set_complex(true);
01785 //      Region filt_region;
01786 
01787 //      if (nx & 1) {
01788 //              LOGERR("even image xsize only");                throw ImageFormatException("even image xsize only");
01789 //      }
01790 
01791         int cs = (((nx * 7 / 4) & 0xfffff8) - nx) / 2; // this pads the image to 1 3/4 * size with result divis. by 8
01792 
01793         static EMData big_clip;
01794         int big_x = nx+2*cs;
01795         int big_y = ny+2*cs;
01796         int big_z = 1;
01797         if ( nz != 1 ) {
01798                 big_z = nz+2*cs;
01799         }
01800 
01801 
01802         if ( big_clip.get_xsize() != big_x || big_clip.get_ysize() != big_y || big_clip.get_zsize() != big_z ) {
01803                 big_clip.set_size(big_x,big_y,big_z);
01804         }
01805         // It is important to set all newly established pixels around the boundaries to the mean
01806         // If this is not done then the associated rotational alignment routine breaks, in fact
01807         // everythin just goes foo.
01808 
01809         big_clip.to_value(get_edge_mean());
01810 
01811         if (nz != 1) {
01812                 big_clip.insert_clip(this,IntPoint(cs,cs,cs));
01813         } else  {
01814                 big_clip.insert_clip(this,IntPoint(cs,cs,0));
01815         }
01816         
01817         // The filter object is nothing more than a cached high pass filter
01818         // Ultimately it is used an argument to the EMData::mult(EMData,prevent_complex_multiplication (bool))
01819         // function in calc_mutual_correlation. Note that in the function the prevent_complex_multiplication
01820         // set to true, which is used for speed reasons.
01821         if (filt->get_xsize() != big_clip.get_xsize() +2-(big_clip.get_xsize()%2) || filt->get_ysize() != big_clip.get_ysize() ||
01822                    filt->get_zsize() != big_clip.get_zsize()) {
01823                 filt->set_size(big_clip.get_xsize() + 2-(big_clip.get_xsize()%2), big_clip.get_ysize(), big_clip.get_zsize());
01824                 filt->to_one();
01825                 filt->process_inplace("filter.highpass.gauss", Dict("cutoff_abs", 1.5f/nx));
01826 #ifdef EMAN2_USING_CUDA
01827                 if(big_clip.cudarwdata)
01828                 {
01829                         filt->copy_to_cuda(); // since this occurs just once for many images, we don't pay much of a speed pentalty here, and we avoid the hassel of messing with sparx
01830                 }
01831 #endif
01832         }
01833 #ifdef EMAN2_USING_CUDA
01834         if(big_clip.cudarwdata && !filt->cudarwdata)
01835         {
01836                 filt->copy_to_cuda(); // since this occurs just once for many images, we don't pay much of a speed pentalty here, and we avoid the hassel of messing with sparx
01837         }
01838 #endif
01839         
01840         EMData *mc = big_clip.calc_mutual_correlation(&big_clip, true,filt);
01841         mc->sub(mc->get_edge_mean());
01842 
01843         static EMData sml_clip;
01844         int sml_x = nx * 3 / 2;
01845         int sml_y = ny * 3 / 2;
01846         int sml_z = 1;
01847         if ( nz != 1 ) {
01848                 sml_z = nz * 3 / 2;
01849         }
01850 
01851         if ( sml_clip.get_xsize() != sml_x || sml_clip.get_ysize() != sml_y || sml_clip.get_zsize() != sml_z ) {
01852                 sml_clip.set_size(sml_x,sml_y,sml_z);   }
01853         if (nz != 1) {
01854                 sml_clip.insert_clip(mc,IntPoint(-cs+nx/4,-cs+ny/4,-cs+nz/4));
01855         } else {
01856                 sml_clip.insert_clip(mc,IntPoint(-cs+nx/4,-cs+ny/4,0));
01857         }
01858                 
01859         delete mc; mc = 0;
01860         EMData * result = NULL;
01861         
01862         if (nz == 1) {
01863                 if (!unwrap) {
01864 #ifdef EMAN2_USING_CUDA
01865                         if(sml_clip.cudarwdata) throw UnexpectedBehaviorException("shap masking is not yet supported by CUDA");
01866 #endif
01867                         result = sml_clip.process("mask.sharp", Dict("outer_radius", -1, "value", 0));
01868 
01869                 }
01870                 else {
01871                         result = sml_clip.unwrap();
01872                 }
01873         }
01874         else {
01875                 // I am not sure why there is any consideration of non 2D images, but it was here
01876                 // in the first port so I kept when I cleaned this function up (d.woolford)
01877 //              result = clipped_mc;
01878                 result = new EMData(sml_clip);
01879         }
01880         
01881 #ifdef EMAN2_USING_CUDA
01882         sml_clip.roneedsanupdate(); //If we didn't do this then unwrap would use data from the previous call of this function, happens b/c sml_clip is static
01883 #endif
01884         EXITFUNC;
01885         if ( unwrap == true)
01886         { // this if statement reflects a strict policy of caching in only one scenario see comments at beginning of function block
01887 
01888                 // Note that the if statement at the beginning of this function ensures that rot_fp is not zero, so there is no need
01889                 // to throw any exception
01890                 if ( rot_fp != 0 ) throw UnexpectedBehaviorException("The rotational foot print is only expected to be cached if it is not NULL");
01891 
01892                 // Here is where the caching occurs - the rot_fp takes ownsherhip of the pointer, and a deep copied EMData object is returned.
01893                 // The deep copy invokes a cost in terms of CPU cycles and memory, but prevents the need for complicated memory management (reference counting)
01894                 rot_fp = result;
01895                 return new EMData(*rot_fp);
01896         }
01897         else return result;
01898 }
01899 
01900 EMData *EMData::make_footprint(int type)
01901 {
01902 //      printf("Make fp %d\n",type);
01903         if (type==0) {
01904                 EMData *un=make_rotational_footprint_e1(); // Use EMAN1's footprint strategy
01905                 if (un->get_ysize() <= 6) {
01906                         throw UnexpectedBehaviorException("In EMData::make_footprint. The rotational footprint is too small");
01907                 }
01908                 EMData *tmp=un->get_clip(Region(0,4,un->get_xsize(),un->get_ysize()-6));        // 4 and 6 are empirical
01909                 EMData *cx=tmp->calc_ccfx(tmp,0,-1,1);
01910                 EMData *fp=cx->get_clip(Region(0,0,cx->get_xsize()/2,cx->get_ysize()));
01911                 delete un;
01912                 delete tmp;
01913                 delete cx;
01914                 return fp;
01915         }
01916         else if (type==1 || type==2 ||type==5 || type==6) {
01917                 int i,j,kx,ky,lx,ly;
01918 
01919                 EMData *fft=do_fft();
01920 
01921                 // map for x,y -> radius for speed
01922                 int rmax=(get_xsize()+1)/2;
01923                 float *rmap=(float *)malloc(rmax*rmax*sizeof(float));
01924                 for (i=0; i<rmax; i++) {
01925                         for (j=0; j<rmax; j++) {
01926 #ifdef _WIN32
01927                                 rmap[i+j*rmax]=_hypotf((float)i,(float)j);
01928 #else
01929                                 rmap[i+j*rmax]=hypot((float)i,(float)j);
01930 #endif  //_WIN32
01931 //                              printf("%d\t%d\t%f\n",i,j,rmap[i+j*rmax]);
01932                         }
01933                 }
01934 
01935                 EMData *fp=new EMData(rmax*2+2,rmax*2,1);
01936                 fp->set_complex(1);
01937                 fp->to_zero();
01938 
01939                 // Two vectors in to complex space (kx,ky) and (lx,ly)
01940                 // We are computing the bispectrum, f(k).f(l).f*(k+l)
01941                 // but integrating out two dimensions, leaving |k|,|l|
01942                 for (kx=-rmax+1; kx<rmax; kx++) {
01943                         for (ky=-rmax+1; ky<rmax; ky++) {
01944                                 for (lx=-rmax+1; lx<rmax; lx++) {
01945                                         for (ly=-rmax+1; ly<rmax; ly++) {
01946                                                 int ax=kx+lx;
01947                                                 int ay=ky+ly;
01948                                                 if (abs(ax)>=rmax || abs(ay)>=rmax) continue;
01949                                                 int r1=(int)floor(.5+rmap[abs(kx)+rmax*abs(ky)]);
01950                                                 int r2=(int)floor(.5+rmap[abs(lx)+rmax*abs(ly)]);
01951 //                                              if (r1>500 ||r2>500) printf("%d\t%d\t%d\t%d\t%d\t%d\n",kx,ky,lx,ly,r1,r2);
01952 //                                              float r3=rmap[ax+rmax*ay];
01953                                                 if (r1+r2>=rmax) continue;
01954 
01955                                                 std::complex<float> p=fft->get_complex_at(kx,ky)*fft->get_complex_at(lx,ly)*conj(fft->get_complex_at(ax,ay));
01956                                                 fp->set_value_at(r1*2,r2,p.real()+fp->get_value_at(r1*2,r2));           // We keep only the real component in anticipation of zero phase sum
01957 //                                              fp->set_value_at(r1*2,rmax*2-r2-1,  fp->get_value_at(r1*2,r2));         // We keep only the real component in anticipation of zero phase sum
01958 //                                              fp->set_value_at(r1*2+1,r2,p.real()+fp->get_value_at(r1*2+1,r2));               // We keep only the real component in anticipation of zero phase sum
01959                                                 fp->set_value_at(r1*2+1,r2,fp->get_value_at(r1*2+1,r2)+1);                      // a normalization counter
01960                                         }
01961                                 }
01962                         }
01963                 }
01964 
01965                 // Normalizes the pixels based on the accumulated counts then sets the imaginary components back to zero
01966                 if (type==5 || type==6) {
01967                         for (i=0; i<rmax*2; i+=2) {
01968                                 for (j=0; j<rmax; j++) {
01969                                         float norm=fp->get_value_at(i+1,j);
01970 #ifdef _WIN32
01971                                         fp->set_value_at(i,rmax*2-j-1,pow(fp->get_value_at(i,j)/(norm==0.0f?1.0f:norm), 1.0f/3.0f));
01972                                         fp->set_value_at(i,j,pow(fp->get_value_at(i,j)/(norm==0.0f?1.0f:norm), 1.0f/3.0f));
01973 #else
01974                                         fp->set_value_at(i,rmax*2-j-1,cbrt(fp->get_value_at(i,j)/(norm==0?1.0:norm)));
01975                                         fp->set_value_at(i,j,cbrt(fp->get_value_at(i,j)/(norm==0?1.0:norm)));
01976 #endif  //_WIN32
01977                                         fp->set_value_at(i+1,j,0.0);
01978                                 }
01979                         }
01980                 }
01981                 else {
01982                         for (i=0; i<rmax*2; i+=2) {
01983                                 for (j=0; j<rmax; j++) {
01984                                         float norm=fp->get_value_at(i+1,j);
01985                                         fp->set_value_at(i,rmax*2-j-1,fp->get_value_at(i,j)/(norm==0.0f?1.0f:norm));
01986                                         fp->set_value_at(i,j,fp->get_value_at(i,j)/(norm==0.0f?1.0f:norm));
01987                                         fp->set_value_at(i+1,j,0.0);
01988                                 }
01989                         }
01990                 }
01991 
01992                 free(rmap);
01993                 if (type==2||type==6) {
01994                         EMData *f2=fp->do_ift();
01995                         if (f2->get_value_at(0,0)<0) f2->mult(-1.0f);
01996                         f2->process_inplace("xform.phaseorigin.tocorner");
01997                         delete fp;
01998                         return f2;
01999                 }
02000                 return fp;
02001         }
02002         else if (type==3 || type==4) {
02003                 int h,i,j,kx,ky,lx,ly;
02004 
02005                 EMData *fft=do_fft();
02006 
02007                 // map for x,y -> radius for speed
02008                 int rmax=(get_xsize()+1)/2;
02009                 float *rmap=(float *)malloc(rmax*rmax*sizeof(float));
02010                 for (i=0; i<rmax; i++) {
02011                         for (j=0; j<rmax; j++) {
02012 #ifdef _WIN32
02013                                 rmap[i+j*rmax]=_hypotf((float)i,(float)j);
02014 #else
02015                                 rmap[i+j*rmax]=hypot((float)i,(float)j);
02016 #endif  //_WIN32
02017 //                              printf("%d\t%d\t%f\n",i,j,rmap[i+j*rmax]);
02018                         }
02019                 }
02020 
02021                 EMData *fp=new EMData(rmax*2+2,rmax*2,16);
02022 
02023                 fp->set_complex(1);
02024                 fp->to_zero();
02025 
02026                 // Two vectors in to complex space (kx,ky) and (lx,ly)
02027                 // We are computing the bispectrum, f(k).f(l).f*(k+l)
02028                 // but integrating out two dimensions, leaving |k|,|l|
02029                 for (kx=-rmax+1; kx<rmax; kx++) {
02030                         for (ky=-rmax+1; ky<rmax; ky++) {
02031                                 for (lx=-rmax+1; lx<rmax; lx++) {
02032                                         for (ly=-rmax+1; ly<rmax; ly++) {
02033                                                 int ax=kx+lx;
02034                                                 int ay=ky+ly;
02035                                                 if (abs(ax)>=rmax || abs(ay)>=rmax) continue;
02036                                                 float rr1=rmap[abs(kx)+rmax*abs(ky)];
02037                                                 float rr2=rmap[abs(lx)+rmax*abs(ly)];
02038                                                 int r1=(int)floor(.5+rr1);
02039                                                 int r2=(int)floor(.5+rr2);
02040 //                                              if (r1>500 ||r2>500) printf("%d\t%d\t%d\t%d\t%d\t%d\n",kx,ky,lx,ly,r1,r2);
02041 //                                              float r3=rmap[ax+rmax*ay];
02042                                                 if (r1+r2>=rmax || rr1==0 ||rr2==0) continue;
02043 
02044                                                 std::complex<float> p=fft->get_complex_at(kx,ky)*fft->get_complex_at(lx,ly)*conj(fft->get_complex_at(ax,ay));
02045                                                 int dot=(int)floor((kx*lx+ky*ly)/(rr1*rr2)*7.5);                                        // projection of k on l 0-31
02046                                                 if (dot<0) dot=16+dot;
02047 //                                              int dot=(int)floor((kx*lx+ky*ly)/(rr1*rr2)*7.5+8.0);                                    // projection of k on l 0-15
02048                                                 fp->set_value_at(r1*2,r2,dot,p.real()+fp->get_value_at(r1*2,r2,dot));           // We keep only the real component in anticipation of zero phase sum
02049 //                                              fp->set_value_at(r1*2,rmax*2-r2-1,  fp->get_value_at(r1*2,r2));         // We keep only the real component in anticipation of zero phase sum
02050 //                                              fp->set_value_at(r1*2+1,r2,p.real()+fp->get_value_at(r1*2+1,r2));               // We keep only the real component in anticipation of zero phase sum
02051                                                 fp->set_value_at(r1*2+1,r2,dot,fp->get_value_at(r1*2+1,r2,dot)+1);                      // a normalization counter
02052                                         }
02053                                 }
02054                         }
02055                 }
02056 
02057                 // Normalizes the pixels based on the accumulated counts then sets the imaginary components back to zero
02058                 for (i=0; i<rmax*2; i+=2) {
02059                         for (j=0; j<rmax; j++) {
02060                                 for (h=0; h<16; h++) {
02061                                         float norm=fp->get_value_at(i+1,j,h);
02062 //                                      fp->set_value_at(i,rmax*2-j-1,h,cbrt(fp->get_value_at(i,j,h)/(norm==0?1.0:norm)));
02063 //                                      fp->set_value_at(i,j,h,cbrt(fp->get_value_at(i,j,h)/(norm==0?1.0:norm)));
02064                                         fp->set_value_at(i,rmax*2-j-1,h,(fp->get_value_at(i,j,h)/(norm==0.0f?1.0f:norm)));
02065                                         fp->set_value_at(i,j,h,(fp->get_value_at(i,j,h)/(norm==0.0f?1.0f:norm)));
02066         //                              fp->set_value_at(i,rmax*2-j-1,fp->get_value_at(i,j)/(norm==0?1.0:norm));
02067         //                              fp->set_value_at(i,j,fp->get_value_at(i,j)/(norm==0?1.0:norm));
02068                                         fp->set_value_at(i+1,j,h,0.0);
02069                                 }
02070                         }
02071                 }
02072 
02073                 free(rmap);
02074                 if (type==4) {
02075                         EMData *f2=fp->do_ift();
02076                         if (f2->get_value_at(0,0,0)<0) f2->mult(-1.0f);
02077                         f2->process_inplace("xform.phaseorigin.tocorner");
02078                         delete fp;
02079                         return f2;
02080                 }
02081                 return fp;
02082         }
02083         throw UnexpectedBehaviorException("There is not implementation for the parameters you specified");
02084 }
02085 
02086 
02087 EMData *EMData::calc_mutual_correlation(EMData * with, bool tocenter, EMData * filter)
02088 {
02089         ENTERFUNC;
02090 
02091         if (with && !EMUtil::is_same_size(this, with)) {
02092                 LOGERR("images not same size");
02093                 throw ImageFormatException( "images not same size");
02094         }
02095 
02096 #ifdef EMAN2_USING_CUDA
02097         if(cudarwdata && with->cudarwdata)
02098         {       
02099 
02100                 EMData* this_fft = do_fft_cuda();
02101 
02102                 EMData *cf = 0;
02103                 if (with && with != this) {
02104                         cf = with->do_fft_cuda();
02105                 }else{
02106                         cf = this_fft->copy();
02107                 }
02108                 
02109                 if (filter) {
02110                         if (!EMUtil::is_same_size(filter, cf)) {
02111                                 LOGERR("improperly sized filter");
02112                                 throw ImageFormatException("improperly sized filter");
02113                         }
02114                         mult_complex_efficient_cuda(cf->cudarwdata, filter->cudarwdata, cf->get_xsize(), cf->get_ysize(), cf->get_zsize(), 1);
02115                         mult_complex_efficient_cuda(this_fft->cudarwdata, filter->cudarwdata, this_fft->get_xsize(), this_fft->get_ysize(), this_fft->get_zsize(), 1);
02116                 }
02117                 
02118                 mcf_cuda(this_fft->cudarwdata, cf->cudarwdata, this_fft->get_xsize(), this_fft->get_ysize(), this_fft->get_zsize());
02119                 
02120                 EMData *f2 = cf->do_ift_cuda();
02121 
02122                 if (tocenter) {
02123                         f2->process_inplace("xform.phaseorigin.tocenter");
02124                 }
02125 
02126                 if( cf )
02127                 {
02128                         delete cf;
02129                         cf = 0;
02130                 }
02131 
02132                 if( this_fft )
02133                 {
02134                         delete this_fft;
02135                         this_fft = 0;
02136                 }
02137 
02138                 f2->set_attr("label", "MCF");
02139                 f2->set_path("/tmp/eman.mcf");
02140 
02141                 EXITFUNC;
02142                 return f2;
02143         }
02144 #endif
02145 
02146         EMData *this_fft = 0;
02147         this_fft = do_fft();
02148 
02149         if (!this_fft) {
02150 
02151                 LOGERR("FFT returns NULL image");
02152                 throw NullPointerException("FFT returns NULL image");
02153         }
02154 
02155         this_fft->ap2ri(); //this is not needed!
02156         EMData *cf = 0;
02157 
02158         if (with && with != this) {
02159                 cf = with->do_fft();
02160                 if (!cf) {
02161                         LOGERR("FFT returns NULL image");
02162                         throw NullPointerException("FFT returns NULL image");
02163                 }
02164                 cf->ap2ri(); //nor is this!
02165         }
02166         else {
02167                 cf = this_fft->copy();
02168         }
02169         
02170         if (filter) {
02171                 if (!EMUtil::is_same_size(filter, cf)) {
02172                         LOGERR("improperly sized filter");
02173                         throw ImageFormatException("improperly sized filter");
02174                 }
02175                 
02176                 cf->mult_complex_efficient(*filter,true); //insanely this is required....
02177                 this_fft->mult(*filter,true);
02178                 //cf->mult_complex_efficient(*filter,7); // takes advantage of the fact that the filter is 1 everywhere but near the origin
02179                 //this_fft->mult_complex_efficient(*filter,7);
02180                 /*cf->mult_complex_efficient(*filter,5);
02181                 this_fft->mult_complex_efficient(*filter,5);*/
02182         }
02183 
02184         float *rdata1 = this_fft->get_data();
02185         float *rdata2 = cf->get_data();
02186         size_t this_fft_size = (size_t)this_fft->get_xsize() * this_fft->get_ysize() * this_fft->get_zsize();
02187 
02188         if (with == this) {
02189                 for (size_t i = 0; i < this_fft_size; i += 2) {
02190                         rdata2[i] = std::sqrt(rdata1[i] * rdata2[i] + rdata1[i + 1] * rdata2[i + 1]);
02191                         rdata2[i + 1] = 0;
02192                 }
02193 
02194                 this_fft->update();
02195                 cf->update();
02196         }
02197         else {
02198                 for (size_t i = 0; i < this_fft_size; i += 2) {
02199                         rdata2[i] = (rdata1[i] * rdata2[i] + rdata1[i + 1] * rdata2[i + 1]);
02200                         rdata2[i + 1] = (rdata1[i + 1] * rdata2[i] - rdata1[i] * rdata2[i + 1]);
02201                 }
02202                 
02203                 //This seems like a bug, but it probably is never used....
02204                 for (size_t i = 0; i < this_fft_size; i += 2) {
02205                         float t = Util::square(rdata2[i]) + Util::square(rdata2[i + 1]);
02206                         if (t != 0) {
02207                                 t = pow(t, 0.25f);
02208                                 rdata2[i] /= t;
02209                                 rdata2[i + 1] /= t;
02210                         }
02211                 }
02212                 this_fft->update();
02213                 cf->update();
02214         }
02215 
02216         EMData *f2 = cf->do_ift();
02217 
02218         if (tocenter) {
02219                 f2->process_inplace("xform.phaseorigin.tocenter");
02220         }
02221 
02222         if( cf )
02223         {
02224                 delete cf;
02225                 cf = 0;
02226         }
02227 
02228         if( this_fft )
02229         {
02230                 delete this_fft;
02231                 this_fft = 0;
02232         }
02233 
02234         f2->set_attr("label", "MCF");
02235         f2->set_path("/tmp/eman.mcf");
02236 
02237         EXITFUNC;
02238         return f2;
02239 }
02240 
02241 
02242 vector < float > EMData::calc_hist(int hist_size, float histmin, float histmax,const float& brt, const float& cont)
02243 {
02244         ENTERFUNC;
02245 
02246         static size_t prime[] = { 1, 3, 7, 11, 17, 23, 37, 59, 127, 253, 511 };
02247 
02248         if (histmin == histmax) {
02249                 histmin = get_attr("minimum");
02250                 histmax = get_attr("maximum");
02251         }
02252 
02253         vector <float> hist(hist_size, 0.0);
02254 
02255         int p0 = 0;
02256         int p1 = 0;
02257         size_t size = (size_t)nx * ny * nz;
02258         if (size < 300000) {
02259                 p0 = 0;
02260                 p1 = 0;
02261         }
02262         else if (size < 2000000) {
02263                 p0 = 2;
02264                 p1 = 3;
02265         }
02266         else if (size < 8000000) {
02267                 p0 = 4;
02268                 p1 = 6;
02269         }
02270         else {
02271                 p0 = 7;
02272                 p1 = 9;
02273         }
02274 
02275         if (is_complex() && p0 > 0) {
02276                 p0++;
02277                 p1++;
02278         }
02279 
02280         size_t di = 0;
02281 //      float norm = 0;
02282         size_t n = hist.size();
02283 
02284         float * data = get_data();
02285         for (int k = p0; k <= p1; ++k) {
02286                 if (is_complex()) {
02287                         di = prime[k] * 2;
02288                 }
02289                 else {
02290                         di = prime[k];
02291                 }
02292 
02293 //              norm += (float)size / (float) di;
02294                 float w = (float)n / (histmax - histmin);
02295 
02296                 for(size_t i=0; i<=size-di; i += di) {
02297                         float val;
02298                         if (cont != 1.0f || brt != 0)val = cont*(data[i]+brt);
02299                         else val = data[i];
02300                         int j = Util::round((val - histmin) * w);
02301                         if (j >= 0 && j < (int) n) {
02302                                 hist[j] += 1;
02303                         }
02304                 }
02305         }
02306 /*
02307         for (size_t i = 0; i < hist.size(); ++i) {
02308                 if (norm != 0) {
02309                         hist[i] = hist[i] / norm;
02310                 }
02311         }
02312 */
02313         return hist;
02314 
02315         EXITFUNC;
02316 }
02317 
02318 
02319 
02320 
02321 
02322 vector<float> EMData::calc_az_dist(int n, float a0, float da, float rmin, float rmax)
02323 {
02324         ENTERFUNC;
02325 
02326         a0=a0*M_PI/180.0f;
02327         da=da*M_PI/180.0f;
02328 
02329         if (get_ndim() > 2) {
02330                 throw ImageDimensionException("no 3D image");
02331         }
02332 
02333         float *yc = new float[n];
02334 
02335         vector<float>   vd(n);
02336         for (int i = 0; i < n; i++) {
02337                 yc[i] = 0.00001f;
02338         }
02339 
02340         float * data = get_data();
02341         if (is_complex()) {
02342                 int c = 0;
02343                 for (int y = 0; y < ny; y++) {
02344                         for (int x = 0; x < nx; x += 2, c += 2) {
02345                                 int x1 = x / 2;
02346                                 int y1 = y<ny/2?y:y-ny;
02347                                 float r = (float)Util::hypot_fast(x1,y1);
02348 
02349                                 if (r >= rmin && r <= rmax) {
02350                                         float a = 0;
02351 
02352                                         if (y != ny / 2 || x != 0) {
02353                                                 a = (atan2((float)y1, (float)x1) - a0) / da;
02354                                         }
02355 
02356                                         int i = (int)(floor(a));
02357                                         a -= i;
02358 
02359                                         if (i == 0) {
02360                                                 vd[0] += data[c] * (1.0f - a);
02361                                                 yc[0] += (1.0f - a);
02362                                         }
02363                                         else if (i == n - 1) {
02364                                                 vd[n - 1] += data[c] * a;
02365                                                 yc[n - 1] += a;
02366                                         }
02367                                         else if (i > 0 && i < (n - 1)) {
02368                                                 float h = 0;
02369                                                 if (is_ri()) {
02370 #ifdef  _WIN32
02371                                                         h = (float)_hypot(data[c], data[c + 1]);
02372 #else
02373                                                         h = (float)hypot(data[c], data[c + 1]);
02374 #endif  //_WIN32
02375                                                 }
02376                                                 else {
02377                                                         h = data[c];
02378                                                 }
02379 
02380                                                 vd[i] += h * (1.0f - a);
02381                                                 yc[i] += (1.0f - a);
02382                                                 vd[i + 1] += h * a;
02383                                                 yc[i + 1] += a;
02384                                         }
02385                                 }
02386                         }
02387                 }
02388         }
02389         else {
02390                 int c = 0;
02391                 float half_nx = (nx - 1) / 2.0f;
02392                 float half_ny = (ny - 1) / 2.0f;
02393 
02394                 for (int y = 0; y < ny; y++) {
02395                         for (int x = 0; x < nx; x++, c++) {
02396                                 float y1 = y - half_ny;
02397                                 float x1 = x - half_nx;
02398 #ifdef  _WIN32
02399                                 float r = (float)_hypot(x1, y1);
02400 #else
02401                                 float r = (float)hypot(x1, y1);
02402 #endif
02403 
02404                                 if (r >= rmin && r <= rmax) {
02405                                         float a = 0;
02406                                         if (x1 != 0 || y1 != 0) {
02407                                                 a = atan2(y1, x1);
02408                                                 if (a < 0) {
02409                                                         a += M_PI * 2;
02410                                                 }
02411                                         }
02412 
02413                                         a = (a - a0) / da;
02414                                         int i = static_cast < int >(floor(a));
02415                                         a -= i;
02416 
02417                                         if (i == 0) {
02418                                                 vd[0] += data[c] * (1.0f - a);
02419                                                 yc[0] += (1.0f - a);
02420                                         }
02421                                         else if (i == n - 1) {
02422                                                 vd[n - 1] += data[c] * a;
02423                                                 yc[n - 1] += (a);
02424                                         }
02425                                         else if (i > 0 && i < (n - 1)) {
02426                                                 vd[i] += data[c] * (1.0f - a);
02427                                                 yc[i] += (1.0f - a);
02428                                                 vd[i + 1] += data[c] * a;
02429                                                 yc[i + 1] += a;
02430                                         }
02431                                 }
02432                         }
02433                 }
02434         }
02435 
02436 
02437         for (int i = 0; i < n; i++) {
02438                 vd[i] /= yc[i];
02439         }
02440 
02441         if( yc )
02442         {
02443                 delete[]yc;
02444                 yc = 0;
02445         }
02446 
02447         return vd;
02448 
02449         EXITFUNC;
02450 }
02451 
02452 
02453 EMData *EMData::unwrap(int r1, int r2, int xs, int dx, int dy, bool do360, bool weight_radial) const
02454 {
02455         ENTERFUNC;
02456 
02457         if (get_ndim() != 2) {
02458                 throw ImageDimensionException("2D image only");
02459         }
02460 
02461         int p = 1;
02462         if (do360) {
02463                 p = 2;
02464         }
02465 
02466         if (xs < 1) {
02467                 xs = (int) Util::fast_floor(p * M_PI * ny / 4);
02468                 xs -= xs % 8;
02469                 if (xs<=8) xs=16;
02470         }
02471 
02472         if (r1 < 0) {
02473                 r1 = 4;
02474         }
02475 
02476 #ifdef  _WIN32
02477         int rr = ny / 2 - 2 - (int) Util::fast_floor(static_cast<float>(_hypot(dx, dy)));
02478 #else
02479         int rr = ny / 2 - 2 - (int) Util::fast_floor(static_cast<float>(hypot(dx, dy)));
02480 #endif  //_WIN32
02481         rr-=rr%2;
02482         if (r2 <= r1 || r2 > rr) {
02483                 r2 = rr;
02484         }
02485 
02486         if ( (r2-r1) < 0 ) throw UnexpectedBehaviorException("The combination of function the arguments and the image dimensions causes unexpected behavior internally. Use a larger image, or a smaller value of r1, or a combination of both");
02487 
02488 #ifdef EMAN2_USING_CUDA
02489         if (isrodataongpu()){
02490                 //cout << " CUDA unwrap" << endl;
02491                 EMData* result = new EMData(0,0,xs,(r2-r1),1);
02492                 result->rw_alloc();
02493                 bindcudaarrayA(true);
02494                 emdata_unwrap(result->cudarwdata, r1, r2, xs, p, dx, dy, weight_radial, nx, ny);
02495                 unbindcudaarryA();
02496                 return result;
02497         }
02498 #endif
02499 
02500         EMData *ret = new EMData();
02501         ret->set_size(xs, r2 - r1, 1);
02502         const float *const d = get_const_data();
02503         float *dd = ret->get_data();
02504         float pfac = (float)p/(float)xs;
02505 
02506         int nxon2 = nx/2;
02507         int nyon2 = ny/2;
02508         for (int x = 0; x < xs; x++) {
02509                 float ang = x * M_PI * pfac;
02510                 float si = sin(ang);
02511                 float co = cos(ang);
02512 
02513                 for (int y = 0; y < r2 - r1; y++) {
02514                         float ypr1 = (float)y + r1;
02515                         float xx = ypr1 * co + nxon2 + dx;
02516                         float yy = ypr1 * si + nyon2 + dy;
02517 //                      float t = xx - Util::fast_floor(xx);
02518 //                      float u = yy - Util::fast_floor(yy);
02519                         float t = xx - (int)xx;
02520                         float u = yy - (int)yy;
02521 //                      int k = (int) Util::fast_floor(xx) + (int) (Util::fast_floor(yy)) * nx;
02522                         int k = (int) xx + ((int) yy) * nx;
02523                         float val = Util::bilinear_interpolate(d[k], d[k + 1], d[k + nx], d[k + nx+1], t,u);
02524                         if (weight_radial) val *=  ypr1;
02525                         dd[x + y * xs] = val;
02526                 }
02527 
02528         }
02529         ret->update();
02530 
02531         EXITFUNC;
02532         return ret;
02533 }
02534 
02535 // NOTE : x axis is from 0 to 0.5  (Nyquist), and thus properly handles non-square images
02536 // complex only
02537 void EMData::apply_radial_func(float x0, float step, vector < float >array, bool interp)
02538 {
02539         ENTERFUNC;
02540 
02541         if (!is_complex()) throw ImageFormatException("apply_radial_func requires a complex image");
02542 
02543         int n = static_cast < int >(array.size());
02544 
02545         if (n*step>2.0) printf("Warning, apply_radial_func takes x0 and step with respect to Nyquist (0.5)\n");
02546 
02547 //      printf("%f %f %f\n",array[0],array[25],array[50]);
02548 
02549         ap2ri();
02550 
02551         size_t ndims = get_ndim();
02552         float * data = get_data();
02553         if (ndims == 2) {
02554                 int k = 0;
02555                 for (int j = 0; j < ny; j++) {
02556                         for (int i = 0; i < nx; i += 2, k += 2) {
02557                                 float r;
02558 #ifdef  _WIN32
02559                                 if (j<ny/2) r = (float)_hypot(i/(float)(nx*2), j/(float)ny);
02560                                 else r = (float)_hypot(i/(float)(nx*2), (ny-j)/(float)ny);
02561 #else
02562                                 if (j<ny/2) r = (float)hypot(i/(float)(nx*2), j/(float)ny);
02563                                 else r = (float)hypot(i/(float)(nx*2), (ny-j)/(float)ny);
02564 #endif  //_WIN32
02565                                 r = (r - x0) / step;
02566 
02567                                 int l = 0;
02568                                 if (interp) {
02569                                         l = (int) floor(r);
02570                                 }
02571                                 else {
02572                                         l = (int) floor(r + 1);
02573                                 }
02574 
02575 
02576                                 float f = 0;
02577                                 if (l >= n - 2) {
02578                                         f = array[n - 1];
02579                                 }
02580                                 else {
02581                                         if (interp) {
02582                                                 r -= l;
02583                                                 f = (array[l] * (1.0f - r) + array[l + 1] * r);
02584                                         }
02585                                         else {
02586                                                 f = array[l];
02587                                         }
02588                                 }
02589 
02590                                 data[k] *= f;
02591                                 data[k + 1] *= f;
02592                         }
02593                 }
02594         }
02595         else if (ndims == 3) {
02596                 int k = 0;
02597                 for (int m = 0; m < nz; m++) {
02598                         float mnz;
02599                         if (m<nz/2) mnz=m*m/(float)(nz*nz);
02600                         else { mnz=(nz-m)/(float)nz; mnz*=mnz; }
02601 
02602                         for (int j = 0; j < ny; j++) {
02603                                 float jny;
02604                                 if (j<ny/2) jny= j*j/(float)(ny*ny);
02605                                 else { jny=(ny-j)/(float)ny; jny*=jny; }
02606 
02607                                 for (int i = 0; i < nx; i += 2, k += 2) {
02608                                         float r = std::sqrt((i * i / (nx*nx*4.0f)) + jny + mnz);
02609                                         r = (r - x0) / step;
02610 
02611                                         int l = 0;
02612                                         if (interp) {
02613                                                 l = (int) floor(r);
02614                                         }
02615                                         else {
02616                                                 l = (int) floor(r + 1);
02617                                         }
02618 
02619 
02620                                         float f = 0;
02621                                         if (l >= n - 2) {
02622                                                 f = array[n - 1];
02623                                         }
02624                                         else {
02625                                                 if (interp) {
02626                                                         r -= l;
02627                                                         f = (array[l] * (1.0f - r) + array[l + 1] * r);
02628                                                 }
02629                                                 else {
02630                                                         f = array[l];
02631                                                 }
02632                                         }
02633 
02634                                         data[k] *= f;
02635                                         data[k + 1] *= f;
02636                                 }
02637                         }
02638                 }
02639 
02640         }
02641 
02642         update();
02643         EXITFUNC;
02644 }
02645 
02646 vector<float> EMData::calc_radial_dist(int n, float x0, float dx, bool inten)
02647 {
02648         ENTERFUNC;
02649 
02650         vector<float>ret(n);
02651         vector<float>norm(n);
02652 
02653         int x,y,z,i;
02654         int step=is_complex()?2:1;
02655         int isinten=get_attr_default("is_intensity",0);
02656 
02657         if (isinten&&!inten) { throw InvalidParameterException("Must set inten for calc_radial_dist with intensity image"); }
02658 
02659         for (i=0; i<n; i++) ret[i]=norm[i]=0.0;
02660         float * data = get_data();
02661 
02662         // We do 2D separately to avoid the hypot3 call
02663         if (nz==1) {
02664                 for (y=i=0; y<ny; y++) {
02665                         for (x=0; x<nx; x+=step,i+=step) {
02666                                 float r,v;
02667                                 if (step==2) {          //complex
02668                                         if (x==0 && y>ny/2) continue;
02669                                         r=(float)(Util::hypot_fast(x/2,y<ny/2?y:ny-y));         // origin at 0,0; periodic
02670                                         if (!inten) {
02671 #ifdef  _WIN32
02672                                                 if (is_ri()) v=static_cast<float>(_hypot(data[i],data[i+1]));   // real/imag, compute amplitude
02673 #else
02674                                                 if (is_ri()) v=static_cast<float>(hypot(data[i],data[i+1]));    // real/imag, compute amplitude
02675 #endif
02676                                                 else v=data[i];                                                 // amp/phase, just get amp
02677                                         } else {
02678                                                 if (isinten) v=data[i];
02679                                                 else if (is_ri()) v=data[i]*data[i]+data[i+1]*data[i+1];
02680                                                 else v=data[i]*data[i];
02681                                         }
02682                                 }
02683                                 else {
02684                                         r=(float)(Util::hypot_fast(x-nx/2,y-ny/2));
02685                                         if (inten) v=data[i]*data[i];
02686                                         else v=data[i];
02687                                 }
02688                                 r=(r-x0)/dx;
02689                                 int f=int(r);   // safe truncation, so floor isn't needed
02690                                 r-=float(f);    // r is now the fractional spacing between bins
02691 //                              printf("%d\t%d\t%d\t%1.3f\t%d\t%1.3f\t%1.4g\n",x,y,f,r,step,Util::hypot_fast(x/2,y<ny/2?y:ny-y),v);
02692                                 if (f>=0 && f<n) {
02693                                         ret[f]+=v*(1.0f-r);
02694                                         norm[f]+=(1.0f-r);
02695                                         if (f<n-1) {
02696                                                 ret[f+1]+=v*r;
02697                                                 norm[f+1]+=r;
02698                                         }
02699                                 }
02700                         }
02701                 }
02702         }
02703         else {
02704                 size_t i;       //3D file may have >2G size
02705                 for (z=i=0; z<nz; ++z) {
02706                         for (y=0; y<ny; ++y) {
02707                                 for (x=0; x<nx; x+=step,i+=step) {
02708                                         float r,v;
02709                                         if (step==2) {  //complex
02710                                                 if (x==0 && z<nz/2) continue;
02711                                                 if (x==0 && z==nz/2 && y<ny/2) continue;
02712                                                 r=Util::hypot3(x/2,y<ny/2?y:ny-y,z<nz/2?z:nz-z);        // origin at 0,0; periodic
02713                                                 if (!inten) {
02714 #ifdef  _WIN32
02715                                                         if (is_ri()) v=static_cast<float>(_hypot(data[i],data[i+1]));   // real/imag, compute amplitude
02716 #else
02717                                                         if (is_ri()) v=static_cast<float>(hypot(data[i],data[i+1]));    // real/imag, compute amplitude
02718 #endif  //_WIN32
02719                                                         else v=data[i];                                                 // amp/phase, just get amp
02720                                                 } else {
02721                                                         if (isinten) v=data[i];
02722                                                         else if (is_ri()) v=data[i]*data[i]+data[i+1]*data[i+1];
02723                                                         else v=data[i]*data[i];
02724                                                 }
02725                                         }
02726                                         else {
02727                                                 r=Util::hypot3(x-nx/2,y-ny/2,z-nz/2);
02728                                                 if (inten) v=data[i]*data[i];
02729                                                 else v=data[i];
02730                                         }
02731                                         r=(r-x0)/dx;
02732                                         int f=int(r);   // safe truncation, so floor isn't needed
02733                                         r-=float(f);    // r is now the fractional spacing between bins
02734                                         if (f>=0 && f<n) {
02735                                                 ret[f]+=v*(1.0f-r);
02736                                                 norm[f]+=(1.0f-r);
02737                                                 if (f<n-1) {
02738                                                         ret[f+1]+=v*r;
02739                                                         norm[f+1]+=r;
02740                                                 }
02741                                         }
02742                                 }
02743                         }
02744                 }
02745         }
02746 
02747         for (i=0; i<n; i++) ret[i]/=norm[i]?norm[i]:1.0f;       // Normalize
02748 
02749         EXITFUNC;
02750 
02751         return ret;
02752 }
02753 
02754 vector<float> EMData::calc_radial_dist(int n, float x0, float dx, int nwedge, bool inten)
02755 {
02756         ENTERFUNC;
02757 
02758         if (nz > 1) {
02759                 LOGERR("2D images only.");
02760                 throw ImageDimensionException("2D images only");
02761         }
02762 
02763         vector<float>ret(n*nwedge);
02764         vector<float>norm(n*nwedge);
02765 
02766         int x,y,i;
02767         int step=is_complex()?2:1;
02768         float astep=static_cast<float>(M_PI*2.0/nwedge);
02769         float* data = get_data();
02770         for (i=0; i<n*nwedge; i++) ret[i]=norm[i]=0.0;
02771 
02772         // We do 2D separately to avoid the hypot3 call
02773         for (y=i=0; y<ny; y++) {
02774                 for (x=0; x<nx; x+=step,i+=step) {
02775                         float r,v,a;
02776                         if (is_complex()) {
02777 #ifdef  _WIN32
02778                                 r=static_cast<float>(_hypot(x/2.0,y<ny/2?y:ny-y));              // origin at 0,0; periodic
02779 #else
02780                                 r=static_cast<float>(hypot(x/2.0,y<ny/2?y:ny-y));               // origin at 0,0; periodic
02781 #endif
02782                                 a=atan2(float(y<ny/2?y:ny-y),x/2.0f);
02783                                 if (!inten) {
02784 #ifdef  _WIN32
02785                                         if (is_ri()) v=static_cast<float>(_hypot(data[i],data[i+1]));   // real/imag, compute amplitude
02786 #else
02787                                         if (is_ri()) v=static_cast<float>(hypot(data[i],data[i+1]));    // real/imag, compute amplitude
02788 #endif  //_WIN32
02789                                         else v=data[i];                                                 // amp/phase, just get amp
02790                                 } else {
02791                                         if (is_ri()) v=data[i]*data[i]+data[i+1]*data[i+1];
02792                                         else v=data[i]*data[i];
02793                                 }
02794                         }
02795                         else {
02796 #ifdef  _WIN32
02797                                 r=static_cast<float>(_hypot(x-nx/2,y-ny/2));
02798 #else
02799                                 r=static_cast<float>(hypot(x-nx/2,y-ny/2));
02800 #endif  //_WIN32
02801                                 a=atan2(float(y-ny/2),float(x-nx/2));
02802                                 if (inten) v=data[i]*data[i];
02803                                 else v=data[i];
02804                         }
02805                         int bin=n*int((a+M_PI)/astep);
02806                         if (bin>=nwedge) bin=nwedge-1;
02807                         r=(r-x0)/dx;
02808                         int f=int(r);   // safe truncation, so floor isn't needed
02809                         r-=float(f);    // r is now the fractional spacing between bins
02810                         if (f>=0 && f<n) {
02811                                 ret[f+bin]+=v*(1.0f-r);
02812                                 norm[f+bin]+=(1.0f-r);
02813                                 if (f<n-1) {
02814                                         ret[f+1+bin]+=v*r;
02815                                         norm[f+1+bin]+=r;
02816                                 }
02817                         }
02818                 }
02819         }
02820 
02821         for (i=0; i<n*nwedge; i++) ret[i]/=norm[i]?norm[i]:1.0f;        // Normalize
02822         EXITFUNC;
02823 
02824         return ret;
02825 }
02826 
02827 void EMData::cconj() {
02828         ENTERFUNC;
02829         if (!is_complex() || !is_ri())
02830                 throw ImageFormatException("EMData::conj requires a complex, ri image");
02831         int nxreal = nx -2 + int(is_fftodd());
02832         int nxhalf = nxreal/2;
02833         for (int iz = 0; iz < nz; iz++)
02834                 for (int iy = 0; iy < ny; iy++)
02835                         for (int ix = 0; ix <= nxhalf; ix++)
02836                                 cmplx(ix,iy,iz) = conj(cmplx(ix,iy,iz));
02837         EXITFUNC;
02838 }
02839 
02840 //#ifdef EMAN2_USING_CUDA
02841 //void EMData::update_stat_cuda() const
02842 //{
02843 //      float* stats = update_stats_cuda(cudarwdata);
02844         
02845 //      attr_dict["mean"] = stats[0];
02846 //      attr_dict["sigma"] = stats[1];
02847         
02848 //      free(stats);
02849 //}
02850 //#endif
02851 
02852 void EMData::update_stat() const
02853 {
02854         ENTERFUNC;
02855 //      printf("update stat %f %d\n",(float)attr_dict["mean"],flags);
02856         if (!(flags & EMDATA_NEEDUPD))
02857         {
02858                 EXITFUNC;
02859                 return;
02860         }
02861 
02862         float* data = get_data();
02863         float max = -FLT_MAX;
02864         float min = -max;
02865 
02866         double sum = 0;
02867         double square_sum = 0;
02868 
02869         int step = 1;
02870         if (is_complex() && !is_ri()) {
02871                 step = 2;
02872         }
02873 
02874         int n_nonzero = 0;
02875 
02876         size_t size = (size_t)nx*ny*nz;
02877         for (size_t i = 0; i < size; i += step) {
02878                 float v = data[i];
02879         #ifdef _WIN32
02880                 max = _cpp_max(max,v);
02881                 min = _cpp_min(min,v);
02882         #else
02883                 max=std::max<float>(max,v);
02884                 min=std::min<float>(min,v);
02885         #endif  //_WIN32
02886                 sum += v;
02887                 square_sum += v * (double)(v);
02888                 if (v != 0) n_nonzero++;
02889         }
02890 
02891         size_t n = size / step;
02892         double mean = sum / n;
02893 
02894 #ifdef _WIN32
02895         float sigma = (float)std::sqrt( _cpp_max(0.0,(square_sum - sum*sum / n)/(n-1)));
02896         n_nonzero = _cpp_max(1,n_nonzero);
02897         double sigma_nonzero = std::sqrt( _cpp_max(0,(square_sum  - sum*sum/n_nonzero)/(n_nonzero-1)));
02898 #else
02899         float sigma = (float)std::sqrt(std::max<double>(0.0,(square_sum - sum*sum / n)/(n-1)));
02900         n_nonzero = std::max<int>(1,n_nonzero);
02901         double sigma_nonzero = std::sqrt(std::max<double>(0,(square_sum  - sum*sum/n_nonzero)/(n_nonzero-1)));
02902 #endif  //_WIN32
02903         double mean_nonzero = sum / n_nonzero; // previous version overcounted! G2
02904 
02905         attr_dict["minimum"] = min;
02906         attr_dict["maximum"] = max;
02907         attr_dict["mean"] = (float)(mean);
02908         attr_dict["sigma"] = (float)(sigma);
02909         attr_dict["square_sum"] = (float)(square_sum);
02910         attr_dict["mean_nonzero"] = (float)(mean_nonzero);
02911         attr_dict["sigma_nonzero"] = (float)(sigma_nonzero);
02912         attr_dict["is_complex"] = (int) is_complex();
02913         attr_dict["is_complex_ri"] = (int) is_ri();
02914 
02915         flags &= ~EMDATA_NEEDUPD;
02916 
02917         if (rot_fp != 0)
02918         {
02919                 delete rot_fp; rot_fp = 0;
02920         }
02921 
02922         EXITFUNC;
02923 //      printf("done stat %f %f %f\n",(float)mean,(float)max,(float)sigma);
02924 }
02925 
02926 bool EMData::operator==(const EMData& that) const {
02927         if (that.get_xsize() != nx || that.get_ysize() != ny || that.get_zsize() != nz ) return false;
02928 
02929         const float*  d1 = that.get_const_data();
02930         float* d2 = get_data();
02931 
02932         for(size_t i =0; i < get_size(); ++i,++d1,++d2) {
02933                 if ((*d1) != (*d2)) return false;
02934         }
02935         return true;
02936 
02937 }
02938 
02939 EMData * EMAN::operator+(const EMData & em, float n)
02940 {
02941         EMData * r = em.copy();
02942         r->add(n);
02943         return r;
02944 }
02945 
02946 EMData * EMAN::operator-(const EMData & em, float n)
02947 {
02948         EMData* r = em.copy();
02949         r->sub(n);
02950         return r;
02951 }
02952 
02953 EMData * EMAN::operator*(const EMData & em, float n)
02954 {
02955         EMData* r = em.copy();
02956         r ->mult(n);
02957         return r;
02958 }
02959 
02960 EMData * EMAN::operator/(const EMData & em, float n)
02961 {
02962         EMData * r = em.copy();
02963         r->div(n);
02964         return r;
02965 }
02966 
02967 
02968 EMData * EMAN::operator+(float n, const EMData & em)
02969 {
02970         EMData * r = em.copy();
02971         r->add(n);
02972         return r;
02973 }
02974 
02975 EMData * EMAN::operator-(float n, const EMData & em)
02976 {
02977         EMData * r = em.copy();
02978         r->mult(-1.0f);
02979         r->add(n);
02980         return r;
02981 }
02982 
02983 EMData * EMAN::operator*(float n, const EMData & em)
02984 {
02985         EMData * r = em.copy();
02986         r->mult(n);
02987         return r;
02988 }
02989 
02990 EMData * EMAN::operator/(float n, const EMData & em)
02991 {
02992         EMData * r = em.copy();
02993         r->to_one();
02994         r->mult(n);
02995         r->div(em);
02996 
02997         return r;
02998 }
02999 
03000 EMData * EMAN::rsub(const EMData & em, float n)
03001 {
03002         return EMAN::operator-(n, em);
03003 }
03004 
03005 EMData * EMAN::rdiv(const EMData & em, float n)
03006 {
03007         return EMAN::operator/(n, em);
03008 }
03009 
03010 EMData * EMAN::operator+(const EMData & a, const EMData & b)
03011 {
03012         EMData * r = a.copy();
03013         r->add(b);
03014         return r;
03015 }
03016 
03017 EMData * EMAN::operator-(const EMData & a, const EMData & b)
03018 {
03019         EMData * r = a.copy();
03020         r->sub(b);
03021         return r;
03022 }
03023 
03024 EMData * EMAN::operator*(const EMData & a, const EMData & b)
03025 {
03026         EMData * r = a.copy();
03027         r->mult(b);
03028         return r;
03029 }
03030 
03031 EMData * EMAN::operator/(const EMData & a, const EMData & b)
03032 {
03033         EMData * r = a.copy();
03034         r->div(b);
03035         return r;
03036 }
03037 
03038 void EMData::set_xyz_origin(float origin_x, float origin_y, float origin_z)
03039 {
03040         attr_dict["origin_x"] = origin_x;
03041         attr_dict["origin_y"] = origin_y;
03042         attr_dict["origin_z"] = origin_z;
03043 }
03044 
03045 #if 0
03046 void EMData::calc_rcf(EMData * with, vector < float >&sum_array)
03047 {
03048         ENTERFUNC;
03049 
03050         int array_size = sum_array.size();
03051         float da = 2 * M_PI / array_size;
03052         float *dat = new float[array_size + 2];
03053         float *dat2 = new float[array_size + 2];
03054         int nx2 = nx * 9 / 20;
03055 
03056         float lim = 0;
03057         if (fabs(translation[0]) < fabs(translation[1])) {
03058                 lim = fabs(translation[1]);
03059         }
03060         else {
03061                 lim = fabs(translation[0]);
03062         }
03063 
03064         nx2 -= static_cast < int >(floor(lim));
03065 
03066         for (int i = 0; i < array_size; i++) {
03067                 sum_array[i] = 0;
03068         }
03069 
03070         float sigma = attr_dict["sigma"];
03071         float with_sigma = with->get_attr_dict().get("sigma");
03072 
03073         vector<float> vdata, vdata2;
03074         for (int i = 8; i < nx2; i += 6) {
03075                 vdata = calc_az_dist(array_size, 0, da, i, i + 6);
03076                 vdata2 = with->calc_az_dist(array_size, 0, da, i, i + 6);
03077                 Assert(vdata.size() <= array_size + 2);
03078                 Assert(cdata2.size() <= array_size + 2);
03079                 std::copy(vdata.begin(), vdata.end(), dat);
03080                 std::copy(vdata2.begin(), vdata2.end(), dat2);
03081 
03082                 EMfft::real_to_complex_1d(dat, dat, array_size);
03083                 EMfft::real_to_complex_1d(dat2, dat2, array_size);
03084 
03085                 for (int j = 0; j < array_size + 2; j += 2) {
03086                         float max = dat[j] * dat2[j] + dat[j + 1] * dat2[j + 1];
03087                         float max2 = dat[j + 1] * dat2[j] - dat2[j + 1] * dat[j];
03088                         dat[j] = max;
03089                         dat[j + 1] = max2;
03090                 }
03091 
03092                 EMfft::complex_to_real_1d(dat, dat, array_size);
03093                 float norm = array_size * array_size * (4.0f * sigma) * (4.0f * with_sigma);
03094 
03095                 for (int j = 0; j < array_size; j++) {
03096                         sum_array[j] += dat[j] * (float) i / norm;
03097                 }
03098         }
03099 
03100         if( dat )
03101         {
03102                 delete[]dat;
03103                 dat = 0;
03104         }
03105 
03106         if( dat2 )
03107         {
03108                 delete[]dat2;
03109                 dat2 = 0;
03110         }
03111         EXITFUNC;
03112 }
03113 
03114 #endif
03115 
03116 void EMData::add_incoherent(EMData * obj)
03117 {
03118         ENTERFUNC;
03119 
03120         if (!obj) {
03121                 LOGERR("NULL image");
03122                 throw NullPointerException("NULL image");
03123         }
03124 
03125         if (!obj->is_complex() || !is_complex()) {
03126                 throw ImageFormatException("complex images only");
03127         }
03128 
03129         if (!EMUtil::is_same_size(this, obj)) {
03130                 throw ImageFormatException("images not same size");
03131         }
03132 
03133         ri2ap();
03134         obj->ri2ap();
03135 
03136         float *dest = get_data();
03137         float *src = obj->get_data();
03138         size_t size = (size_t)nx * ny * nz;
03139         for (size_t j = 0; j < size; j += 2) {
03140 #ifdef  _WIN32
03141                 dest[j] = (float) _hypot(src[j], dest[j]);
03142 #else
03143                 dest[j] = (float) hypot(src[j], dest[j]);
03144 #endif  //_WIN32
03145                 dest[j + 1] = 0;
03146         }
03147 
03148         obj->update();
03149         update();
03150         EXITFUNC;
03151 }
03152 
03153 
03154 float EMData::calc_dist(EMData * second_img, int y_index) const
03155 {
03156         ENTERFUNC;
03157 
03158         if (get_ndim() != 1) {
03159                 throw ImageDimensionException("'this' image is 1D only");
03160         }
03161 
03162         if (second_img->get_xsize() != nx || ny != 1) {
03163                 throw ImageFormatException("image xsize not same");
03164         }
03165 
03166         if (y_index > second_img->get_ysize() || y_index < 0) {
03167                 return -1;
03168         }
03169 
03170         float ret = 0;
03171         float *d1 = get_data();
03172         float *d2 = second_img->get_data() + second_img->get_xsize() * y_index;
03173 
03174         for (int i = 0; i < nx; i++) {
03175                 ret += Util::square(d1[i] - d2[i]);
03176         }
03177         EXITFUNC;
03178         return std::sqrt(ret);
03179 }
03180 
03181 
03182 EMData * EMData::calc_fast_sigma_image( EMData* mask)
03183 {
03184         ENTERFUNC;
03185 
03186         bool maskflag = false;
03187         if (mask == 0) {
03188                 mask = new EMData(nx,ny,nz);
03189                 mask->process_inplace("testimage.circlesphere");
03190                 maskflag = true;
03191         }
03192 
03193         if (get_ndim() != mask->get_ndim() ) throw ImageDimensionException("The dimensions do not match");
03194 
03195         int mnx = mask->get_xsize(); int mny = mask->get_ysize(); int mnz = mask->get_zsize();
03196 
03197         if ( mnx > nx || mny > ny || mnz > nz)
03198                 throw ImageDimensionException("Can not calculate variance map using an image that is larger than this image");
03199 
03200         size_t P = 0;
03201         for(size_t i = 0; i < mask->get_size(); ++i){
03202                 if (mask->get_value_at(i) != 0){
03203                         ++P;
03204                 }
03205         }
03206         float normfac = 1.0f/(float)P;
03207 
03208 //      bool undoclip = false;
03209 
03210         int nxc = nx+mnx; int nyc = ny+mny; int nzc = nz+mnz;
03211 //      if ( mnx < nx || mny < ny || mnz < nz) {
03212         Region r;
03213         if (ny == 1) r = Region((mnx-nxc)/2,nxc);
03214         else if (nz == 1) r = Region((mnx-nxc)/2, (mny-nyc)/2,nxc,nyc);
03215         else r = Region((mnx-nxc)/2, (mny-nyc)/2,(mnz-nzc)/2,nxc,nyc,nzc);
03216         mask->clip_inplace(r,0.0);
03217         //Region r((mnx-nxc)/2, (mny-nyc)/2,(mnz-nzc)/2,nxc,nyc,nzc);
03218         //mask->clip_inplace(r);
03219         //undoclip = true;
03220         //}
03221 
03222         // Here we generate the local average of the squares
03223         Region r2;
03224         if (ny == 1) r2 = Region((nx-nxc)/2,nxc);
03225         else if (nz == 1) r2 = Region((nx-nxc)/2, (ny-nyc)/2,nxc,nyc);
03226         else r2 = Region((nx-nxc)/2, (ny-nyc)/2,(nz-nzc)/2,nxc,nyc,nzc);
03227         EMData* squared = get_clip(r2,get_edge_mean());
03228 
03229         EMData* tmp = squared->copy();
03230         Dict pow;
03231         pow["pow"] = 2.0f;
03232         squared->process_inplace("math.pow",pow);
03233         EMData* s = mask->convolute(squared);//ming, mask squared exchange
03234         squared->mult(normfac);
03235 
03236         EMData* m = mask->convolute(tmp);//ming, tmp mask exchange
03237         m->mult(normfac);
03238         m->process_inplace("math.pow",pow);
03239         delete tmp; tmp = 0;
03240         s->sub(*m);
03241         // Here we finally generate the standard deviation image
03242         s->process_inplace("math.sqrt");
03243 
03244 //      if ( undoclip ) {
03245 //              Region r((nx-mnx)/2, (ny-mny)/2, (nz-mnz)/2,mnx,mny,mnz);
03246 //              mask->clip_inplace(r);
03247 //      }
03248 
03249         if (maskflag) {
03250                 delete mask;
03251                 mask = 0;
03252         } else {
03253                 Region r;
03254                 if (ny == 1) r = Region((nxc-mnx)/2,mnx);
03255                 else if (nz == 1) r = Region((nxc-mnx)/2, (nyc-mny)/2,mnx,mny);
03256                 else r = Region((nxc-mnx)/2, (nyc-mny)/2,(nzc-mnz)/2,mnx,mny,mnz);
03257                 mask->clip_inplace(r);
03258         }
03259 
03260         delete squared;
03261         delete m;
03262 
03263         s->process_inplace("xform.phaseorigin.tocenter");
03264         Region r3;
03265         if (ny == 1) r3 = Region((nxc-nx)/2,nx);
03266         else if (nz == 1) r3 = Region((nxc-nx)/2, (nyc-ny)/2,nx,ny);
03267         else r3 = Region((nxc-nx)/2, (nyc-ny)/2,(nzc-nz)/2,nx,ny,nz);
03268         s->clip_inplace(r3);
03269         EXITFUNC;
03270         return s;
03271 }
03272 
03273 //  The following code looks strange - does anybody know it?  Please let me know, pawel.a.penczek@uth.tmc.edu  04/09/06.
03274 // This is just an implementation of "Roseman's" fast normalized cross-correlation (Ultramicroscopy, 2003). But the contents of this function have changed dramatically since you wrote that comment (d.woolford).
03275 EMData *EMData::calc_flcf(EMData * with)
03276 {
03277         ENTERFUNC;
03278         EMData *this_copy=this;
03279         this_copy=copy();
03280 
03281         int mnx = with->get_xsize(); int mny = with->get_ysize(); int mnz = with->get_zsize();
03282         int nxc = nx+mnx; int nyc = ny+mny; int nzc = nz+mnz;
03283 
03284         // Ones is a circular/spherical mask, consisting of 1s.
03285         EMData* ones = new EMData(mnx,mny,mnz);
03286         ones->process_inplace("testimage.circlesphere");
03287 
03288         // Get a copy of with, we will eventually resize it
03289         EMData* with_resized = with->copy();
03290         with_resized->process_inplace("normalize");
03291         with_resized->mult(*ones);
03292 
03293         EMData* s = calc_fast_sigma_image(ones);// Get the local sigma image
03294 
03295         Region r1;
03296         if (ny == 1) r1 = Region((mnx-nxc)/2,nxc);
03297         else if (nz == 1) r1 = Region((mnx-nxc)/2, (mny-nyc)/2,nxc,nyc);
03298         else r1 = Region((mnx-nxc)/2, (mny-nyc)/2,(mnz-nzc)/2,nxc,nyc,nzc);
03299         with_resized->clip_inplace(r1,0.0);
03300 
03301         Region r2;
03302         if (ny == 1) r2 = Region((nx-nxc)/2,nxc);
03303         else if (nz == 1) r2 = Region((nx-nxc)/2, (ny-nyc)/2,nxc,nyc);
03304         else r2 = Region((nx-nxc)/2, (ny-nyc)/2,(nz-nzc)/2,nxc,nyc,nzc);
03305         this_copy->clip_inplace(r2,0.0);
03306 
03307         EMData* corr = this_copy->calc_ccf(with_resized); // the ccf results should have same size as sigma
03308 
03309         corr->process_inplace("xform.phaseorigin.tocenter");
03310         Region r3;
03311         if (ny == 1) r3 = Region((nxc-nx)/2,nx);
03312         else if (nz == 1) r3 = Region((nxc-nx)/2, (nyc-ny)/2,nx,ny);
03313         else r3 = Region((nxc-nx)/2, (nyc-ny)/2,(nzc-nz)/2,nx,ny,nz);
03314         corr->clip_inplace(r3);
03315 
03316         corr->div(*s);
03317 
03318         delete with_resized; delete ones; delete this_copy; delete s;
03319         EXITFUNC;
03320         return corr;
03321 }
03322 
03323 EMData *EMData::convolute(EMData * with)
03324 {
03325         ENTERFUNC;
03326 
03327         EMData *f1 = do_fft();
03328         if (!f1) {
03329                 LOGERR("FFT returns NULL image");
03330                 throw NullPointerException("FFT returns NULL image");
03331         }
03332 
03333         f1->ap2ri();
03334 
03335         EMData *cf = 0;
03336         if (with) {
03337                 cf = with->do_fft();
03338                 if (!cf) {
03339                         LOGERR("FFT returns NULL image");
03340                         throw NullPointerException("FFT returns NULL image");
03341                 }
03342                 cf->ap2ri();
03343         }
03344         else {
03345                 cf = f1->copy();
03346         }
03347         //printf("cf_x=%d, f1y=%d, thisx=%d, withx=%d\n",cf->get_xsize(),f1->get_ysize(),this->get_xsize(),with->get_xsize());
03348         if (with && !EMUtil::is_same_size(f1, cf)) {
03349                 LOGERR("images not same size");
03350                 throw ImageFormatException("images not same size");
03351         }
03352 
03353         float *rdata1 = f1->get_data();
03354         float *rdata2 = cf->get_data();
03355         size_t cf_size = (size_t)cf->get_xsize() * cf->get_ysize() * cf->get_zsize();
03356 
03357         float re,im;
03358 
03359         for (size_t i = 0; i < cf_size; i += 2) {
03360                 re = rdata1[i] * rdata2[i] - rdata1[i + 1] * rdata2[i + 1];
03361                 im = rdata1[i + 1] * rdata2[i] + rdata1[i] * rdata2[i + 1];
03362                 rdata2[i]=re;
03363                 rdata2[i+1]=im;
03364         }
03365         cf->update();
03366         EMData *f2 = cf->do_ift();//ming change cf to cf_temp
03367         //printf("cf_x=%d, f2x=%d, thisx=%d, withx=%d\n",cf->get_xsize(),f2->get_xsize(),this->get_xsize(),with->get_xsize());
03368         if( cf )
03369         {
03370                 delete cf;
03371                 cf = 0;
03372         }
03373 
03374         if( f1 )
03375         {
03376                 delete f1;
03377                 f1=0;
03378         }
03379 
03380         EXITFUNC;
03381         return f2;
03382 }
03383 
03384 
03385 void EMData::common_lines(EMData * image1, EMData * image2,
03386                                                   int mode, int steps, bool horizontal)
03387 {
03388         ENTERFUNC;
03389 
03390         if (!image1 || !image2) {
03391                 throw NullPointerException("NULL image");
03392         }
03393 
03394         if (mode < 0 || mode > 2) {
03395                 throw OutofRangeException(0, 2, mode, "invalid mode");
03396         }
03397 
03398         if (!image1->is_complex()) {
03399                 image1 = image1->do_fft();
03400         }
03401         if (!image2->is_complex()) {
03402                 image2 = image2->do_fft();
03403         }
03404 
03405         image1->ap2ri();
03406         image2->ap2ri();
03407 
03408         if (!EMUtil::is_same_size(image1, image2)) {
03409                 throw ImageFormatException("images not same sizes");
03410         }
03411 
03412         int image2_nx = image2->get_xsize();
03413         int image2_ny = image2->get_ysize();
03414 
03415         int rmax = image2_ny / 4 - 1;
03416         int array_size = steps * rmax * 2;
03417         float *im1 = new float[array_size];
03418         float *im2 = new float[array_size];
03419         for (int i = 0; i < array_size; i++) {
03420                 im1[i] = 0;
03421                 im2[i] = 0;
03422         }
03423 
03424         set_size(steps * 2, steps * 2, 1);
03425 
03426         float *image1_data = image1->get_data();
03427         float *image2_data = image2->get_data();
03428 
03429         float da = M_PI / steps;
03430         float a = -M_PI / 2.0f + da / 2.0f;
03431         int jmax = 0;
03432 
03433         for (int i = 0; i < steps * 2; i += 2, a += da) {
03434                 float s1 = 0;
03435                 float s2 = 0;
03436                 int i2 = i * rmax;
03437                 int j = 0;
03438 
03439                 for (float r = 3.0f; r < rmax - 3.0f; j += 2, r += 1.0f) {
03440                         float x = r * cos(a);
03441                         float y = r * sin(a);
03442 
03443                         if (x < 0) {
03444                                 x = -x;
03445                                 y = -y;
03446                                 LOGERR("CCL ERROR %d, %f !\n", i, -x);
03447                         }
03448 
03449                         int k = (int) (floor(x) * 2 + floor(y + image2_ny / 2) * image2_nx);
03450                         int l = i2 + j;
03451                         float x2 = x - floor(x);
03452                         float y2 = y - floor(y);
03453 
03454                         im1[l] = Util::bilinear_interpolate(image1_data[k],
03455                                                                                                 image1_data[k + 2],
03456                                                                                                 image1_data[k + image2_nx],
03457                                                                                                 image1_data[k + 2 + image2_nx], x2, y2);
03458 
03459                         im2[l] = Util::bilinear_interpolate(image2_data[k],
03460                                                                                                 image2_data[k + 2],
03461                                                                                                 image2_data[k + image2_nx],
03462                                                                                                 image2_data[k + 2 + image2_nx], x2, y2);
03463 
03464                         k++;
03465 
03466                         im1[l + 1] = Util::bilinear_interpolate(image1_data[k],
03467                                                                                                         image1_data[k + 2],
03468                                                                                                         image1_data[k + image2_nx],
03469                                                                                                         image1_data[k + 2 + image2_nx], x2, y2);
03470 
03471                         im2[l + 1] = Util::bilinear_interpolate(image2_data[k],
03472                                                                                                         image2_data[k + 2],
03473                                                                                                         image2_data[k + image2_nx],
03474                                                                                                         image2_data[k + 2 + image2_nx], x2, y2);
03475 
03476                         s1 += Util::square_sum(im1[l], im1[l + 1]);
03477                         s2 += Util::square_sum(im2[l], im2[l + 1]);
03478                 }
03479 
03480                 jmax = j - 1;
03481                 float sqrt_s1 = std::sqrt(s1);
03482                 float sqrt_s2 = std::sqrt(s2);
03483 
03484                 int l = 0;
03485                 for (float r = 1; r < rmax; r += 1.0f) {
03486                         int i3 = i2 + l;
03487                         im1[i3] /= sqrt_s1;
03488                         im1[i3 + 1] /= sqrt_s1;
03489                         im2[i3] /= sqrt_s2;
03490                         im2[i3 + 1] /= sqrt_s2;
03491                         l += 2;
03492                 }
03493         }
03494         float * data = get_data();
03495 
03496         switch (mode) {
03497         case 0:
03498                 for (int m1 = 0; m1 < 2; m1++) {
03499                         for (int m2 = 0; m2 < 2; m2++) {
03500 
03501                                 if (m1 == 0 && m2 == 0) {
03502                                         for (int i = 0; i < steps; i++) {
03503                                                 int i2 = i * rmax * 2;
03504                                                 for (int j = 0; j < steps; j++) {
03505                                                         int l = i + j * steps * 2;
03506                                                         int j2 = j * rmax * 2;
03507                                                         data[l] = 0;
03508                                                         for (int k = 0; k < jmax; k++) {
03509                                                                 data[l] += im1[i2 + k] * im2[j2 + k];
03510                                                         }
03511                                                 }
03512                                         }
03513                                 }
03514                                 else {
03515                                         int steps2 = steps * m2 + steps * steps * 2 * m1;
03516 
03517                                         for (int i = 0; i < steps; i++) {
03518                                                 int i2 = i * rmax * 2;
03519                                                 for (int j = 0; j < steps; j++) {
03520                                                         int j2 = j * rmax * 2;
03521                                                         int l = i + j * steps * 2 + steps2;
03522                                                         data[l] = 0;
03523 
03524                                                         for (int k = 0; k < jmax; k += 2) {
03525                                                                 i2 += k;
03526                                                                 j2 += k;
03527                                                                 data[l] += im1[i2] * im2[j2];
03528                                                                 data[l] += -im1[i2 + 1] * im2[j2 + 1];
03529                                                         }
03530                                                 }
03531                                         }
03532                                 }
03533                         }
03534                 }
03535 
03536                 break;
03537         case 1:
03538                 for (int m1 = 0; m1 < 2; m1++) {
03539                         for (int m2 = 0; m2 < 2; m2++) {
03540                                 int steps2 = steps * m2 + steps * steps * 2 * m1;
03541                                 int p1_sign = 1;
03542                                 if (m1 != m2) {
03543                                         p1_sign = -1;
03544                                 }
03545 
03546                                 for (int i = 0; i < steps; i++) {
03547                                         int i2 = i * rmax * 2;
03548 
03549                                         for (int j = 0; j < steps; j++) {
03550                                                 int j2 = j * rmax * 2;
03551 
03552                                                 int l = i + j * steps * 2 + steps2;
03553                                                 data[l] = 0;
03554                                                 float a = 0;
03555 
03556                                                 for (int k = 0; k < jmax; k += 2) {
03557                                                         i2 += k;
03558                                                         j2 += k;
03559 
03560 #ifdef  _WIN32
03561                                                         float a1 = (float) _hypot(im1[i2], im1[i2 + 1]);
03562 #else
03563                                                         float a1 = (float) hypot(im1[i2], im1[i2 + 1]);
03564 #endif  //_WIN32
03565                                                         float p1 = atan2(im1[i2 + 1], im1[i2]);
03566                                                         float p2 = atan2(im2[j2 + 1], im2[j2]);
03567 
03568                                                         data[l] += Util::angle_sub_2pi(p1_sign * p1, p2) * a1;
03569                                                         a += a1;
03570                                                 }
03571 
03572                                                 data[l] /= (float)(a * M_PI / 180.0f);
03573                                         }
03574                                 }
03575                         }
03576                 }
03577 
03578                 break;
03579         case 2:
03580                 for (int m1 = 0; m1 < 2; m1++) {
03581                         for (int m2 = 0; m2 < 2; m2++) {
03582                                 int steps2 = steps * m2 + steps * steps * 2 * m1;
03583 
03584                                 for (int i = 0; i < steps; i++) {
03585                                         int i2 = i * rmax * 2;
03586 
03587                                         for (int j = 0; j < steps; j++) {
03588                                                 int j2 = j * rmax * 2;
03589                                                 int l = i + j * steps * 2 + steps2;
03590                                                 data[l] = 0;
03591 
03592                                                 for (int k = 0; k < jmax; k += 2) {
03593                                                         i2 += k;
03594                                                         j2 += k;
03595 #ifdef  _WIN32
03596                                                         data[l] += (float) (_hypot(im1[i2], im1[i2 + 1]) * _hypot(im2[j2], im2[j2 + 1]));
03597 #else
03598                                                         data[l] += (float) (hypot(im1[i2], im1[i2 + 1]) * hypot(im2[j2], im2[j2 + 1]));
03599 #endif  //_WIN32
03600                                                 }
03601                                         }
03602                                 }
03603                         }
03604                 }
03605 
03606                 break;
03607         default:
03608                 break;
03609         }
03610 
03611         if (horizontal) {
03612                 float *tmp_array = new float[ny];
03613                 for (int i = 1; i < nx; i++) {
03614                         for (int j = 0; j < ny; j++) {
03615                                 tmp_array[j] = get_value_at(i, j);
03616                         }
03617                         for (int j = 0; j < ny; j++) {
03618                                 set_value_at(i, j, 0, tmp_array[(j + i) % ny]);
03619                         }
03620                 }
03621                 if( tmp_array )
03622                 {
03623                         delete[]tmp_array;
03624                         tmp_array = 0;
03625                 }
03626         }
03627 
03628         if( im1 )
03629         {
03630                 delete[]im1;
03631                 im1 = 0;
03632         }
03633 
03634         if( im2 )
03635         {
03636                 delete im2;
03637                 im2 = 0;
03638         }
03639 
03640 
03641         image1->update();
03642         image2->update();
03643         if( image1 )
03644         {
03645                 delete image1;
03646                 image1 = 0;
03647         }
03648         if( image2 )
03649         {
03650                 delete image2;
03651                 image2 = 0;
03652         }
03653         update();
03654         EXITFUNC;
03655 }
03656 
03657 
03658 
03659 void EMData::common_lines_real(EMData * image1, EMData * image2,
03660                                                            int steps, bool horiz)
03661 {
03662         ENTERFUNC;
03663 
03664         if (!image1 || !image2) {
03665                 throw NullPointerException("NULL image");
03666         }
03667 
03668         if (!EMUtil::is_same_size(image1, image2)) {
03669                 throw ImageFormatException("images not same size");
03670         }
03671 
03672         int steps2 = steps * 2;
03673         int image_ny = image1->get_ysize();
03674         EMData *image1_copy = image1->copy();
03675         EMData *image2_copy = image2->copy();
03676 
03677         float *im1 = new float[steps2 * image_ny];
03678         float *im2 = new float[steps2 * image_ny];
03679 
03680         EMData *images[] = { image1_copy, image2_copy };
03681         float *ims[] = { im1, im2 };
03682 
03683         for (int m = 0; m < 2; m++) {
03684                 float *im = ims[m];
03685                 float a = M_PI / steps2;
03686                 Transform t(Dict("type","2d","alpha",-a));
03687                 for (int i = 0; i < steps2; i++) {
03688                         images[i]->transform(t);
03689                         float *data = images[i]->get_data();
03690 
03691                         for (int j = 0; j < image_ny; j++) {
03692                                 float sum = 0;
03693                                 for (int k = 0; k < image_ny; k++) {
03694                                         sum += data[j * image_ny + k];
03695                                 }
03696                                 im[i * image_ny + j] = sum;
03697                         }
03698 
03699                         float sum1 = 0;
03700                         float sum2 = 0;
03701                         for (int j = 0; j < image_ny; j++) {
03702                                 int l = i * image_ny + j;
03703                                 sum1 += im[l];
03704                                 sum2 += im[l] * im[l];
03705                         }
03706 
03707                         float mean = sum1 / image_ny;
03708                         float sigma = std::sqrt(sum2 / image_ny - sum1 * sum1);
03709 
03710                         for (int j = 0; j < image_ny; j++) {
03711                                 int l = i * image_ny + j;
03712                                 im[l] = (im[l] - mean) / sigma;
03713                         }
03714 
03715                         images[i]->update();
03716                         a += M_PI / steps;
03717                 }
03718         }
03719 
03720         set_size(steps2, steps2, 1);
03721         float *data1 = get_data();
03722 
03723         if (horiz) {
03724                 for (int i = 0; i < steps2; i++) {
03725                         for (int j = 0, l = i; j < steps2; j++, l++) {
03726                                 if (l == steps2) {
03727                                         l = 0;
03728                                 }
03729 
03730                                 float sum = 0;
03731                                 for (int k = 0; k < image_ny; k++) {
03732                                         sum += im1[i * image_ny + k] * im2[l * image_ny + k];
03733                                 }
03734                                 data1[i + j * steps2] = sum;
03735                         }
03736                 }
03737         }
03738         else {
03739                 for (int i = 0; i < steps2; i++) {
03740                         for (int j = 0; j < steps2; j++) {
03741                                 float sum = 0;
03742                                 for (int k = 0; k < image_ny; k++) {
03743                                         sum += im1[i * image_ny + k] * im2[j * image_ny + k];
03744                                 }
03745                                 data1[i + j * steps2] = sum;
03746                         }
03747                 }
03748         }
03749 
03750         update();
03751 
03752         if( image1_copy )
03753         {
03754                 delete image1_copy;
03755                 image1_copy = 0;
03756         }
03757 
03758         if( image2_copy )
03759         {
03760                 delete image2_copy;
03761                 image2_copy = 0;
03762         }
03763 
03764         if( im1 )
03765         {
03766                 delete[]im1;
03767                 im1 = 0;
03768         }
03769 
03770         if( im2 )
03771         {
03772                 delete[]im2;
03773                 im2 = 0;
03774         }
03775         EXITFUNC;
03776 }
03777 
03778 
03779 void EMData::cut_slice(const EMData *const map, const Transform& transform, bool interpolate)
03780 {
03781         ENTERFUNC;
03782 
03783         if (!map) throw NullPointerException("NULL image");
03784         // These restrictions should be ultimately restricted so that all that matters is get_ndim() = (map->get_ndim() -1)
03785         if ( get_ndim() != 2 ) throw ImageDimensionException("Can not call cut slice on an image that is not 2D");
03786         if ( map->get_ndim() != 3 ) throw ImageDimensionException("Can not cut slice from an image that is not 3D");
03787         // Now check for complex images - this is really just being thorough
03788         if ( is_complex() ) throw ImageFormatException("Can not call cut slice on an image that is complex");
03789         if ( map->is_complex() ) throw ImageFormatException("Can not cut slice from a complex image");
03790 
03791 
03792         float *sdata = map->get_data();
03793         float *ddata = get_data();
03794 
03795         int map_nx = map->get_xsize();
03796         int map_ny = map->get_ysize();
03797         int map_nz = map->get_zsize();
03798         int map_nxy = map_nx * map_ny;
03799 
03800         int ymax = ny/2;
03801         if ( ny % 2 == 1 ) ymax += 1;
03802         int xmax = nx/2;
03803         if ( nx % 2 == 1 ) xmax += 1;
03804         for (int y = -ny/2; y < ymax; y++) {
03805                 for (int x = -nx/2; x < xmax; x++) {
03806                         Vec3f coord(x,y,0);
03807                         Vec3f soln = transform*coord;
03808 
03809 //                      float xx = (x+pretrans[0]) * (*ort)[0][0] +  (y+pretrans[1]) * (*ort)[0][1] + pretrans[2] * (*ort)[0][2] + posttrans[0];
03810 //                      float yy = (x+pretrans[0]) * (*ort)[1][0] +  (y+pretrans[1]) * (*ort)[1][1] + pretrans[2] * (*ort)[1][2] + posttrans[1];
03811 //                      float zz = (x+pretrans[0]) * (*ort)[2][0] +  (y+pretrans[1]) * (*ort)[2][1] + pretrans[2] * (*ort)[2][2] + posttrans[2];
03812 
03813 
03814 //                      xx += map_nx/2;
03815 //                      yy += map_ny/2;
03816 //                      zz += map_nz/2;
03817 
03818                         float xx = soln[0]+map_nx/2;
03819                         float yy = soln[1]+map_ny/2;
03820                         float zz = soln[2]+map_nz/2;
03821 
03822                         int l = (x+nx/2) + (y+ny/2) * nx;
03823 
03824                         float t = xx - floor(xx);
03825                         float u = yy - floor(yy);
03826                         float v = zz - floor(zz);
03827 
03828                         if (xx < 0 || yy < 0 || zz < 0 ) {
03829                                 ddata[l] = 0;
03830                                 continue;
03831                         }
03832                         if (interpolate) {
03833                                 if ( xx > map_nx - 1 || yy > map_ny - 1 || zz > map_nz - 1) {
03834                                         ddata[l] = 0;
03835                                         continue;
03836                                 }
03837                                 int k = (int) (Util::fast_floor(xx) + Util::fast_floor(yy) * map_nx + Util::fast_floor(zz) * map_nxy);
03838 
03839 
03840                                 if (xx < (map_nx - 1) && yy < (map_ny - 1) && zz < (map_nz - 1)) {
03841                                         ddata[l] = Util::trilinear_interpolate(sdata[k],
03842                                                                 sdata[k + 1], sdata[k + map_nx],sdata[k + map_nx + 1],
03843                                                                 sdata[k + map_nxy], sdata[k + map_nxy + 1], sdata[k + map_nx + map_nxy],
03844                                                                 sdata[k + map_nx + map_nxy + 1],t, u, v);
03845                                 }
03846                                 else if ( xx == (map_nx - 1) && yy == (map_ny - 1) && zz == (map_nz - 1) ) {
03847                                         ddata[l] += sdata[k];
03848                                 }
03849                                 else if ( xx == (map_nx - 1) && yy == (map_ny - 1) ) {
03850                                         ddata[l] +=     Util::linear_interpolate(sdata[k], sdata[k + map_nxy],v);
03851                                 }
03852                                 else if ( xx == (map_nx - 1) && zz == (map_nz - 1) ) {
03853                                         ddata[l] += Util::linear_interpolate(sdata[k], sdata[k + map_nx],u);
03854                                 }
03855                                 else if ( yy == (map_ny - 1) && zz == (map_nz - 1) ) {
03856                                         ddata[l] += Util::linear_interpolate(sdata[k], sdata[k + 1],t);
03857                                 }
03858                                 else if ( xx == (map_nx - 1) ) {
03859                                         ddata[l] += Util::bilinear_interpolate(sdata[k], sdata[k + map_nx], sdata[k + map_nxy], sdata[k + map_nxy + map_nx],u,v);
03860                                 }
03861                                 else if ( yy == (map_ny - 1) ) {
03862                                         ddata[l] += Util::bilinear_interpolate(sdata[k], sdata[k + 1], sdata[k + map_nxy], sdata[k + map_nxy + 1],t,v);
03863                                 }
03864                                 else if ( zz == (map_nz - 1) ) {
03865                                         ddata[l] += Util::bilinear_interpolate(sdata[k], sdata[k + 1], sdata[k + map_nx], sdata[k + map_nx + 1],t,u);
03866                                 }
03867 
03868 //                              if (k >= map->get_size()) {
03869 //                                      cout << xx << " " << yy << " " <<  zz << " " << endl;
03870 //                                      cout << k << " " << get_size() << endl;
03871 //                                      cout << get_xsize() << " " << get_ysize() << " " << get_zsize() << endl;
03872 //                                      throw;
03873 //                                      }
03874 //
03875 //                              ddata[l] = Util::trilinear_interpolate(sdata[k],
03876 //                                              sdata[k + 1], sdata[k + map_nx],sdata[k + map_nx + 1],
03877 //                                              sdata[k + map_nxy], sdata[k + map_nxy + 1], sdata[k + map_nx + map_nxy],
03878 //                                              sdata[k + map_nx + map_nxy + 1],t, u, v);
03879                         }
03880                         else {
03881                                 if ( xx > map_nx - 1 || yy > map_ny - 1 || zz > map_nz - 1) {
03882                                         ddata[l] = 0;
03883                                         continue;
03884                                 }
03885                                 size_t k = Util::round(xx) + Util::round(yy) * map_nx + Util::round(zz) * (size_t)map_nxy;
03886                                 ddata[l] = sdata[k];
03887                         }
03888 
03889                 }
03890         }
03891 
03892         update();
03893 
03894         EXITFUNC;
03895 }
03896 
03897 EMData *EMData::unwrap_largerR(int r1,int r2,int xs, float rmax_f) {
03898         float *d,*dd;
03899         int do360=2;
03900         int rmax = (int)(rmax_f+0.5f);
03901         unsigned long i;
03902         unsigned int nvox=get_xsize()*get_ysize();//ming
03903         float maxmap=0.0f, minmap=0.0f;
03904         float temp=0.0f, diff_den=0.0f, norm=0.0f;
03905         float cut_off_va =0.0f;
03906 
03907         d=get_data();
03908         maxmap=-1000000.0f;
03909         minmap=1000000.0f;
03910         for (i=0;i<nvox;i++){
03911                 if(d[i]>maxmap) maxmap=d[i];
03912                 if(d[i]<minmap) minmap=d[i];
03913         }
03914         diff_den = maxmap-minmap;
03915         for(i=0;i<nvox;i++) {
03916                 temp = (d[i]-minmap)/diff_den;
03917                 if(cut_off_va != 0.0) {               // cut off the lowerset ?% noisy information
03918                         if(temp < cut_off_va)
03919                                 d[i] = 0.0f;                   // set the empty part density=0.0
03920                         else
03921                                 d[i] = temp-cut_off_va;
03922                 }
03923                 else    d[i] = temp;
03924         }
03925 
03926         for(i=0;i<nvox;i++) {
03927                 temp=d[i];
03928                 norm += temp*temp;
03929         }
03930         for(i=0;i<nvox;i++)             d[i] /= norm;                      //  y' = y/norm(y)
03931 
03932         if (xs<1) {
03933                 xs = (int) floor(do360*M_PI*get_ysize()/4); // ming
03934                 xs=Util::calc_best_fft_size(xs); // ming
03935         }
03936         if (r1<0) r1=0;
03937         float maxext=ceil(0.6f*std::sqrt((float)(get_xsize()*get_xsize()+get_ysize()*get_ysize())));// ming add std::
03938 
03939         if (r2<r1) r2=(int)maxext;
03940         EMData *ret = new EMData;
03941 
03942         ret->set_size(xs,r2+1,1);
03943 
03944         dd=ret->get_data();
03945 
03946         for (int i=0; i<xs; i++) {
03947                 float si=sin(i*M_PI*2/xs);
03948                 float co=cos(i*M_PI*2/xs);
03949                 for (int r=0; r<=maxext; r++) {
03950                         float x=(r+r1)*co+get_xsize()/2; // ming
03951                         float y=(r+r1)*si+get_ysize()/2; // ming
03952                         if(x<0.0 || x>=get_xsize()-1.0 || y<0.0 || y>=get_ysize()-1.0 || r>rmax){    //Ming , ~~~~ rmax need pass here
03953                                 for(;r<=r2;r++)                                   // here r2=MAXR
03954                                         dd[i+r*xs]=0.0;
03955                         break;
03956                     }
03957                         int x1=(int)floor(x);
03958                         int y1=(int)floor(y);
03959                         float t=x-x1;
03960                         float u=y-y1;
03961                         float f11= d[x1+y1*get_xsize()]; // ming
03962                         float f21= d[(x1+1)+y1*get_xsize()]; // ming
03963                         float f12= d[x1+(y1+1)*get_xsize()]; // ming
03964                         float f22= d[(x1+1)+(y1+1)*get_xsize()]; // ming
03965                         dd[i+r*xs] = (1-t)*(1-u)*f11+t*(1-u)*f21+t*u*f22+(1-t)*u*f12;
03966                 }
03967         }
03968         update();
03969         ret->update();
03970         return ret;
03971 }
03972 
03973 
03974 EMData *EMData::oneDfftPolar(int size, float rmax, float MAXR){         // sent MAXR value here later!!
03975         float *pcs=get_data();
03976         EMData *imagepcsfft = new EMData;
03977         imagepcsfft->set_size((size+2), (int)MAXR+1, 1);
03978         float *d=imagepcsfft->get_data();
03979 
03980         EMData *data_in=new EMData;
03981         data_in->set_size(size,1,1);
03982         float *in=data_in->get_data();
03983 
03984         for(int row=0; row<=(int)MAXR; ++row){
03985                 if(row<=(int)rmax) {
03986                         for(int i=0; i<size;++i)        in[i] = pcs[i+row*size]; // ming
03987                         data_in->set_complex(false);
03988                         data_in->do_fft_inplace();
03989                         for(int j=0;j<size+2;j++)  d[j+row*(size+2)]=in[j];
03990                 }
03991                 else for(int j=0;j<size+2;j++) d[j+row*(size+2)]=0.0;
03992         }
03993         imagepcsfft->update();
03994         delete data_in;
03995         return imagepcsfft;
03996 }
03997 
03998 
03999 
04000 //#ifdef EMAN2_USING_CUDA
04001 //EMData* EMData::cut_slice_cuda(const Transform& transform)
04002 //{
04003 //      ENTERFUNC;
04004 //
04005         // These restrictions should be ultimately restricted so that all that matters is get_ndim() = (map->get_ndim() -1)
04006 //      if ( get_ndim() != 3 ) throw ImageDimensionException("Can not cut slice from an image that is not 3D");
04007         // Now check for complex images - this is really just being thorough
04008 //      if ( is_complex() ) throw ImageFormatException("Can not call cut slice an image that is complex");
04009 //
04010 
04011 //      EMData* ret = new EMData();
04012 //      ret->set_size_cuda(nx,ny,1);
04013 
04014 //      float * m = new float[12];
04015 //      transform.copy_matrix_into_array(m);
04016 
04017 //      EMDataForCuda tmp = ret->get_data_struct_for_cuda();
04018 //      bind_cuda_texture(); // Binds this image to the global texture
04019 //      cut_slice_cuda_(&tmp,m);
04020 
04021 //      ret->gpu_update();
04022 //      delete [] m;
04023 
04024 //      EXITFUNC;
04025 //      return ret;
04026 //}
04027 
04028 //#endif // EMAN2_USING_CUDA
04029 
04030 
04031 void EMData::uncut_slice(EMData * const map, const Transform& transform) const
04032 {
04033         ENTERFUNC;
04034 
04035         if (!map) throw NullPointerException("NULL image");
04036         // These restrictions should be ultimately restricted so that all that matters is get_ndim() = (map->get_ndim() -1)
04037         if ( get_ndim() != 2 ) throw ImageDimensionException("Can not call cut slice on an image that is not 2D");
04038         if ( map->get_ndim() != 3 ) throw ImageDimensionException("Can not cut slice from an image that is not 3D");
04039         // Now check for complex images - this is really just being thorough
04040         if ( is_complex() ) throw ImageFormatException("Can not call cut slice on an image that is complex");
04041         if ( map->is_complex() ) throw ImageFormatException("Can not cut slice from a complex image");
04042 
04043 //      Transform3D r( 0, 0, 0); // EMAN by default
04044 //      if (!ort) {
04045 //              ort = &r;
04046 //      }
04047 
04048         float *ddata = map->get_data();
04049         float *sdata = get_data();
04050 
04051         int map_nx = map->get_xsize();
04052         int map_ny = map->get_ysize();
04053         int map_nz = map->get_zsize();
04054         int map_nxy = map_nx * map_ny;
04055         float map_nz_round_limit = (float) map_nz-0.5f;
04056         float map_ny_round_limit = (float) map_ny-0.5f;
04057         float map_nx_round_limit = (float) map_nx-0.5f;
04058 /*
04059         Vec3f posttrans = ort->get_posttrans();
04060         Vec3f pretrans = ort->get_pretrans();*/
04061 
04062         int ymax = ny/2;
04063         if ( ny % 2 == 1 ) ymax += 1;
04064         int xmax = nx/2;
04065         if ( nx % 2 == 1 ) xmax += 1;
04066         for (int y = -ny/2; y < ymax; y++) {
04067                 for (int x = -nx/2; x < xmax; x++) {
04068                         Vec3f coord(x,y,0);
04069                         Vec3f soln = transform*coord;
04070 //                      float xx = (x+pretrans[0]) * (*ort)[0][0] +  (y+pretrans[1]) * (*ort)[0][1] + pretrans[2] * (*ort)[0][2] + posttrans[0];
04071 //                      float yy = (x+pretrans[0]) * (*ort)[1][0] +  (y+pretrans[1]) * (*ort)[1][1] + pretrans[2] * (*ort)[1][2] + posttrans[1];
04072 //                      float zz = (x+pretrans[0]) * (*ort)[2][0] +  (y+pretrans[1]) * (*ort)[2][1] + pretrans[2] * (*ort)[2][2] + posttrans[2];
04073 //
04074 //                      xx += map_nx/2;
04075 //                      yy += map_ny/2;
04076 //                      zz += map_nz/2;
04077 //
04078                         float xx = soln[0]+map_nx/2;
04079                         float yy = soln[1]+map_ny/2;
04080                         float zz = soln[2]+map_nz/2;
04081 
04082                         // These 0.5 offsets are here because the round function rounds to the nearest whole number.
04083                         if (xx < -0.5 || yy < -0.5 || zz < -0.5 || xx >= map_nx_round_limit || yy >= map_ny_round_limit || zz >= map_nz_round_limit) continue;
04084 
04085                         size_t k = Util::round(xx) + Util::round(yy) * map_nx + Util::round(zz) * (size_t)map_nxy;
04086                         int l = (x+nx/2) + (y+ny/2) * nx;
04087                         ddata[k] = sdata[l];
04088                 }
04089         }
04090 
04091         map->update();
04092         EXITFUNC;
04093 }
04094 
04095 
04096 void EMData::save_byteorder_to_dict(ImageIO * imageio)
04097 {
04098         string image_endian = "ImageEndian";
04099         string host_endian = "HostEndian";
04100 
04101         if (imageio->is_image_big_endian()) {
04102                 attr_dict[image_endian] = "big";
04103         }
04104         else {
04105                 attr_dict[image_endian] = "little";
04106         }
04107 
04108         if (ByteOrder::is_host_big_endian()) {
04109                 attr_dict[host_endian] = "big";
04110         }
04111         else {
04112                 attr_dict[host_endian] = "little";
04113         }
04114 }
04115 

Generated on Mon Mar 7 18:15:25 2011 for EMAN2 by  doxygen 1.3.9.1