Re: [scipion-users] xmipp3 - correlation alignment - Advanced parameters

[Carlos O. S. <co...@cn...>]

Dear Dmitry,


On 02/15/17 14:04, Dmitry Semchonok wrote:
> Dear colleagues,
>
> I have a question about *xmipp3 - correlation alignment.*
>
> In Advanced option if I change to
>
> *Filter at* 2.5 A (instead of 4)
This filter is used to downsample the images resulting in a speed-up and 
removal of noise (just for the record, the alignment operation is 
performed at a sampling rate of newTs=0.9*maxFreq/2, meaning that this 
maximum frequency is mapped to the digital frequency 0.45, Nyquist is at 
0.5). Remind that this program is simply looking for a shift between 
every pair of frames, and then combining all that information in a 
single shift per frame. In this sense, I would doubt that very high 
frequencies result in a significant change of the shift (the spectrum up 
to 4A should be informative enough to be able to detect 2 numbers, X and 
Y shift).
>
> *Maximum shift (pixels) *from 30 to 50
The maxShift is used to search in the autocorrelation function for the 
maximum cross correlation peak between -maxShift to +maxShift. This is 
the maxShift between any two pairs of frames. If you expect larger 
shifts, then you may certainly increase this value. If you don't expect 
these large shifts, then by increasing the maxShift parameter you are 
increasing the probability of identifying a spurious correlation due to 
noise as the largest correlation peak, resulting in an incorrect 
estimate of the shift between the two frames (still, the algorithm 
combining all shifts between pairs in a shift for the frame explicitly 
looks for outliers, so that you noisy identification still may be harmless).

I hope I had clarified how the algorithm works, Carlos Oscar
>
> Will the script work more accurate and precise?
>
> Thank you!
>
> Sincerely,
> Dmitry
>
>
> ------------------------------------------------------------------------------
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, SlashDot.org! http://sdm.link/slashdot
>
>
> _______________________________________________
> scipion-users mailing list
> sci...@li...
> https://lists.sourceforge.net/lists/listinfo/scipion-users

-- 
------------------------------------------------------------------------
Carlos Oscar Sánchez Sorzano                  e-mail:   co...@cn...
Biocomputing unit                             http://biocomp.cnb.csic.es
National Center of Biotechnology (CSIC)
c/Darwin, 3
Campus Universidad Autónoma (Cantoblanco)     Tlf: 34-91-585 4510
28049 MADRID (SPAIN)                          Fax: 34-91-585 4506
------------------------------------------------------------------------