autoTrim in starChooser still not great

Question

autoTrim in starChooser still not great

Mikea1985 opened this issue 7 years ago · 4 comments

Mikea1985 commented 7 years ago

I put forward that autoTrim (in psfStarChooser.starChooser) is still not identifying stars reliably:

Answer 1 · 2019-08-07T01:18:27.000Z

My latest experience with autoTrim is vastly superior to all prior experiences. From V1.0 it seems actually useful. :-) I have yet to see it do anything terrible; at most it has marked a few stars (out of several dozen) as ignored where I think "what's wrong with those?"

Answer 2 · 2019-08-07T17:01:53.000Z

That's fantastic. Really glad my efforts weren't in vain! ha Next on the list is to include the clustering I was talking about. Specifically, I have found if one clusters on the 4D space of FWHM, A, B, and Kron-aperture magnitude, then stars are clustered very cleanly in the cluster that has the smallest mean FWHM, if one starts with SNR>~150 sources. This only really works when one has a moderate to large FOV, or a dense field. But it works beautifully over 10 TB worth of HSC data! :)

…

On Tue, Aug 6, 2019 at 6:18 PM Mike Alexandersen ***@***.***> wrote: My latest experience with autoTrim is vastly superior to all prior experiences. From V1.0 it seems actually useful. :-) I have yet to see it do anything terrible; at most it has marked a few stars (out of several dozen) as ignored where I think "what's wrong with those?" — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#21?email_source=notifications&email_token=ABRJZPXX2OGFT6KU34IIZXDQDIPGHA5CNFSM4FB7ZA62YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD3W4YZQ#issuecomment-518900838>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABRJZPVTM45ZXWLGLSLINQLQDIPGHANCNFSM4FB7ZA6Q> .

Answer 3 · 2019-08-08T01:49:06.000Z

One method for identifying stars that I learnt about from somewhere one the Pan-STARRS catalog website (or somewhere else talking about it) is to compare the Kron and small aperture photometry: for stars the two methods should result in very similar mags, while for galaxies the small aperture underestimates the total flux. Maybe that can be useful here too? I know SExtractor does several kinds of photometry.

Answer 4 · 2019-08-08T03:14:07.000Z

That's one of the parameters I cluster on. Does very well for all but spheroidal galaxies, which the others can trim out.

…

On Wed., Aug. 7, 2019, 6:49 p.m. Mike Alexandersen ***@***.***> wrote: One method for identifying stars that I learnt about from somewhere one the Pan-STARRS catalog website (or somewhere else talking about it) is to compare the Kron and small aperture photometry: for stars the two methods should result in very similar mags, while for galaxies the small aperture underestimates the total flux. Maybe that can be useful here too? I know SExtractor does several kinds of photometry. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#21?email_source=notifications&email_token=ABRJZPWMHBRBLRTK5F5CYA3QDN3RFA5CNFSM4FB7ZA62YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD32FXJY#issuecomment-519330727>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABRJZPQMMAHB7ZOTET5LO5DQDN3RFANCNFSM4FB7ZA6Q> .