basisfilter

This takes an orthogonal basis set as created by svdcmp and uses it to operate on a set of images

basisfilter <input> <basis set> [mask=<rad>] [frac=<num>/<denom>] [nbasis=<n>] [nbasiscls=<n>] [clsbylcmp] [filter=<output>] [classify=<ref set>] [logit=<label>] [align=<ref set>] [multiali=<n>] [groups=<ngrp>] [poweralign[=<max dx>]] [footprint=<fp out>] [bias=<image>] [precen] [adaptive] [basis0+] [saveali=<output>] [origrt] [scalebasis] [verbose] [quiet]

Parameters:


<input>Input images to process
<basis set>A set of orthogonal basis images as generated by svdcmp
[mask=<rad>]Mask radius
[frac=<num>/<denom>]Operate only on a fraction of the raw images
[nbasis=<n>]Number of basis images to use from the basis set. Defaults to all.
[nbasiscls=<n>]Number of basis images to use for classification. Defaults to nbasis.
[clsbylcmp]Classify using 'lcmp', linear comparison in real space (without basis projection)
[filter=<output>]Uses the basis set as a filter on the data
[classify=<ref set>]Performs classification of the input images, <ref set> should be projections to use for classification.
[logit=<label>]This specifies a label to use in the particle.log file. This records particle number classification throughout a refinement.
[align=<ref set>]Align input images to the best match from <ref set>, should be same set used to generate <basis set>
[multiali=<n>]Rather than just taking the best 'align' match, this will take the several best matches and select the one with the strongest decomposition
[groups=<ngrp>]This will classify into multiple sets. Used by svdmultirefine.py
[poweralign[=<max dx>]]Align particles by projected power maximization
[footprint=<fp out>]This will write the decomposition vectors into a footprint file
[bias=<image>]Each input particle will be multiplied by <image> before further processing
[precen]Rotational alignment only
[adaptive]Will try to optimize the number of basis images for each particle.
[basis0+]Insures that mean value of 1st basis vector is positive
[saveali=<output>]Saves aligned unfiltered particles in <output>
[origrt]If the inputs are aligned before filtration, this will cause the filtered output to be returned to the original orienatation
[scalebasis]Scale the basis vectors by their significance in <ref set>
[verbose]More verbose output
[quiet]No logging

Usage:

basisfilter start.hed msa.hed start.filt.hed align=proj.hed origrt

Description

This program performs various operations on a set of (noisy) input images using an orthogonal basis set. The basis vectors are generated by singlar value decomposition/multivariate statistical analysis/prinicpal component analysis (effectively different terms for the same thing in this context). Generally this set is produced using the program 'svdcmp'.

When used to filter data, this program will produce images with dramatically less noise, however, somewhat like a low-pass filter, substantial real information will be missing from the filtered images, particularly at high resolution. The input images are generally a set of projections of a 3D model. If this 3D model is substantially different from the model the noisy images derive from, this filtration could induce significant artifacts.

This routine is mainly aimed at providing less noisy particles for use in alignment/orientation determination. Particles filtered through this program should not be used as actual data for reconstruction.


EMAN Manual page, generated Wed Feb 18 10:33:43 2009