imstatics nclip/lower,upper incompatible?

Andy Newsam wrote on Jul 31, 2006

Hello.

I'm having a problem with imstatistics. I am using the "upper" and "lower" keywords to exclude values from the statistics calulations (these are "masked" pixels in fact) but as soon as I set nclip > 0, it seems to ignore upper/lower limts. As an example:

cl> imstatistics test1.fits lower=0 nclip=0    
#               IMAGE      NPIX      MEAN     MIDPT    STDDEV       MIN       MAX
           test1.fits    702714     1.423     1.134     210.3        0.   113034.

[Correct]

cl> imstatistics test1.fits lower=0 nclip=3 lsigma=4 usigma=3
#               IMAGE      NPIX      MEAN     MIDPT    STDDEV       MIN       MAX
           test1.fits   2308317   -0.5979   -0.9952    0.6923       -1.     2.115

[Clearly not setting lower=0!]

I cant find anything in the documentation or a past buglog - can anyone help?

Cheers
Andy

---

Mike Fitzpatrick wrote on Jul 31, 2006

In your first example: nclip=0
In your second example: nclip=3

The 'nclip' parameter controls the number of iterations performed and if non-zero will cause the lower/upper limits to be adjusted.

If you really want to do statistics through masks, see the MIMSTATISTICS task. See also the examples in the help page for IMEXPR for ways to create pixel masks based on pixel value.

Cheers,
-Mike

---

Andy Newsam wrote on Jul 31, 2006

Thanks - that is pretty much what I was already doing, but I wasn't sure why imstatistics wasn't doing what I expected (naively I assumed that the iterative clipping would be done after the lower and upper limits had been applied to the data)

Cheers
Andy

---

Last post on Jul 31, 2006