cenalignint

This iteratively centers a set of particles, using only integer shifts to avoid interpolation artifacts.

cenalignint <input file> [frac=<0 to d-1>/<d>] [mask=<rad>] [maxshift=<pixels>] [usefilt]

Parameters:


<input file>File containing images to center
[frac=<0 to d-1>/<d>]Processes a fraction of the input file.
[mask=<rad>]Applies a mask to the centered images
[maxshift=<pixels>]Maximum shift in pixels
[usefilt]Use .filt file (filtered particles) for centering, but write original images in output

Usage:

cenalignint init.img frac=0,4

cenalignint init.img frac=1,4

cenalignint init.img frac=2,4

cenalignint init.img frac=3,4

Description

Cenalignint is used to take a set of boxed particles and move each particle to the center of its box (+- 1 pixel). This is accomplished by iteratively aligning each particle to a rotationally averaged image which is the average of all of the particles. This process may not work well for all particles. It typically works very well on objects with high symmetry (icosahedral viruses), and seems to work well on many objects with lower symmetry. Note that this program does only integer pixel translations, so no interpolation is done.

This program is not used by the standard refinement proceedure, but can be used manually before the refinement loop for several purposes: 1) After centering the particles it may be possible to reduce the size (in pixels) of the entire particle set. This will reduce the processing time in the refinement process. 2) Some unusable particles can be eliminated so processor time isn't wasted on them later. 3) The refinement loop may work a little better/converge faster if the particles start out near the center of their boxes.

If, during the iterative loop, a particle does not seem to align to a consistent center, that particle is discarded and written to 'bad.hed'. All sucessfully aligned images are written to 'ali.hed'. The reference image used for alignment after each iteration is stored in 'avg.hed'. This file can be examined to see if convergence really has been reached.

This process generally works better if the references, and eventually the particles, can be masked during each step. The mask radius should be specified so the mask is just larger than the largest axis of your particle.


EMAN Manual page, generated Fri Feb 8 16:12:23 2002