• Segment3D

segment3d

• Split a 3d map into several pieces.

Usage

• segment3d <input file> <output file> [watershed[=<lp>[,<hp>]] [nseg=<n seg>] [morph=<vector file>] [ampw] [thr=<val>] [split] [maxit=<n>] [itxp=<n>] [vector=<outfile>] [apix=<apix>] [chimeraout=<outfile.cmm>] [pdb=<outfile>] [gaussmod=<mrc out>] [randomize] [stats] [segmaxv] [resegbystats=<thr>] [sym=<sym>]

Parameters

• <input file>	Source file
  <output file>	Destination file
  [watershed[=<lp>[,<hp>]]	Apply simplistic watershed segmentation instead of the default (k-means), with an optional pre-filter
  [nseg=<n seg>]	Number of segments to split map into
  [morph=<vector file>]	Takes the result of a previous segmentation and uses it to seed a new segmentation
  [ampw]	Apply a straight line integral amplitude weight to the distance parameter
  [thr=<val>]	Only segment voxels above the given threshold, neg values use sigma multipliers
  [split]	Separate out the individual segments
  [maxit=<n>]	Specify maximum iterations before giving up on convergence
  [itxp=<n>]	After the segmentation, iteratively extend the segments into regions below threshold, n is the number of voxels of expansion
  [vector=<outfile>]	Write segment centers to a text file x,y,z,dx,dy,dz
  [apix=<apix>]	A/pix for chimera output
  [chimeraout=<outfile.cmm>]	Write segment centers to a chimera marker file
  [pdb=<outfile>]	Write segment centers to a PDB file
  [gaussmod=<mrc out>]	Construct a 3d model from Gaussians centered at the group centers.
  [randomize]	This will randomize the gaussians by rotating them randomly with a fixed radius
  [stats]	Produce a file containing mean density values in each segment
  [segmaxv]	Instead of an integer segmap, writes the 'max' value for the segment
  [resegbystats=<thr>]	Merge segments with mean densities above or below the threshold value
  [sym=<sym>]	Insure a symmetric segmentation

  Typical usage: segment3d threed.1a.mrc seg.mrc 12 thr=1 Description

Description

This program will take a 3D map and that map into 'segments'. That is, every voxel in the map will be assigned to one of segments. The assignment is performed in such a way that high density regions will be grouped together.

Note that using the 'watershed' option combined with 'nseg=' will not produce a true 'watershed' segmentation of the image. Instead, it will produce nseg segments then will 'flood fill' the remaining areas above the threshold. EMAN Manual page, generated Mon Jan 8 17:35:59 2007