Three methods are employed:
* Differential Phase Residual (DPR)In this method the the resolution is defined as the spatial frequency at which the average phase discrepancy between the two transforms exceeds 45 degrees.
* Fourier Ring Correlation (FRC)In this method the resolution is defined as the spatial frequency at which annular samplings of the two Fourier transforms register negligible cross-correlation.
* Spectral Signal-to-Noise Ratio (SSNR)This method is based on the measurement of the signal-to-noise ratio as a function of spatial frequency. The SSNR is determined by comparing the Fourier transform of individual images with that of the global average image. (this option is only available for 2D images not for volumes, see the program ssnr if you are interested in computing the signal to noise ratio in 3D).
To calculate "the resolution of a reconstruction", use --set_of_images, where the program divides the corresponding set of projection images into two subsets and reconstructs two volumes from these subsets. This program may then be used to calculate the DPR and FRC between these two volumes. The resulting plots are commonly used to assess the high-resolution limit of the initial reconstruction (see Frank's book for more details).
The program writes out filename.frc files, for each input volume or image, or selfilename.frc, the set_of_images mode. These ACSII files contain the DPR, FRC and SSNR as a function of resolution (in 1/Angstrom). The .frc files also contain a column for the FRC expected for pure noise.
See also
resolution_ssnr
Parameters
-i <input_file>
( requires --ref ) --set_of_images <selfile>
--oroot <root_file=>
-o <output_file=>
--ref <input_file>
--sampling_rate, -s <Ts=1>
--dont_apply_geo
--do_dpr
--max_sam <max_sr=-1>
xmipp_resolution_fsc --ref subset1.vol -i subset2.vol --sam 5.6Resolution of a set of images using 5.6 pixel size (in Angstrom):
xmipp_resolution_fsc --set_of_images selfile.sel --sam 5.6