[daip] CNTR (levels)

[Eric Greisen]

Ian Pattison writes:
 > I wish to set CLEV to be equal to the rms locally in an image (since I
 > have ran PBCOR and so the noise varies alot over my wide-field image).
 > What is the most accurate way to measure the noise? I have tried 2 methods
 > so far:
 > 
 > 1) using TVWIN and IMSTAT to find the rms in a number of boxes close to
 > the source of interest.
 > 
 > 2) using IMEAN to find the "true" rms over a larger box covering the
 > source (source at centre).
 > 
 > These two methods provide similar, but not equal, results. How "good" is
 > IMEAN at calculating the true noise rms (since the rms of the source must
 > somehow be subtracted).
 > 

IMEAN is very good at finding the rms - it Gaussian fits a histogram
of values guided by the input guess at the rms and average.  IMSTAT
does a simple rms computation using all values.  If you have a current
31DEC02 IMEAN even sets the adverb PIXSTD.  If not, it also sets a
header keyword which you can fetch with GETHEAD.

Eric Greisen