daofind seems to be ignoring search parameters
bwp wrote on Jun 26, 2007
I am having some difficulties with noao.apphot.daofind. As I understand it, the main parameters that determine how many objects I detect are "sigma" and "fwhm". However, I get the same number of objects using sigma=4 as sigma=15, and the number is ridiculously high. The objects I'm looking at are not precisely point sources, but they should be pretty darn close. Any idea what I might be doing wrong? Thanks.
-Brad Peterson
Iowa State University
-Brad Peterson
Iowa State University
Valeev Azamat wrote on Jun 26, 2007
In help for daofind you may read:
After image convolution , DAOFIND steps through starmap searching for density enhancements greater than findpars.threshold * datapars.sigma, and brighter than all other density enhancements within a semi-major axis of 0.42466 findpars.nsigma * datapars.fwhmpsf. As the program selects candidates, it computes three shape characteristics, sharpness and 2 estimates of roundness.
Not only findpars.nsigma && datapars.fwhmpsf determine
object count. Parameter findpars.threshold is very important.
Try again with findpars.threshold=4 and findpars.threshold=30.
You can see object number changing.
Is this help?
Azamat
After image convolution , DAOFIND steps through starmap searching for density enhancements greater than findpars.threshold * datapars.sigma, and brighter than all other density enhancements within a semi-major axis of 0.42466 findpars.nsigma * datapars.fwhmpsf. As the program selects candidates, it computes three shape characteristics, sharpness and 2 estimates of roundness.
Not only findpars.nsigma && datapars.fwhmpsf determine
object count. Parameter findpars.threshold is very important.
Try again with findpars.threshold=4 and findpars.threshold=30.
You can see object number changing.
Is this help?
Azamat
bwp wrote on Jun 26, 2007
I will try that. Thank you for your help.
-Brad
-Brad
bwp wrote on Jun 26, 2007
This has indeed helped. Part of my problem was that I was not using the datapars.sigma describing the background at all, I was just using findpars.threshold as the sigma threshold for significance.
Thanks again.
-Brad
Thanks again.
-Brad
bwp wrote on Jun 26, 2007
azastro
In help for daofind you may read:
After image convolution , DAOFIND steps through starmap searching for density enhancements greater than findpars.threshold * datapars.sigma, and brighter than all other density enhancements within a semi-major axis of 0.42466 findpars.nsigma * datapars.fwhmpsf.
Followup question: what if my background sigma varies considerably across the image? Do I chop it into sections to run daofind with different values of datapars.sigma? Or should I just get background stats on a low-background part of the image and use that throughout?
-Brad
Jayme Derrah wrote on Jun 26, 2007
I'm having pretty much the same problem, but changing the threshold doesn't change anything, it's still finding way too many stars!
Last post on Jun 26, 2007