Re: [Astrometry] Matching with One Quad Instead of Two

[Dustin Lang]

Hi,

Sorry for the slow reply -- for some reason your message ended up in Spam!

For 20" x 20" frames (assuming you do mean 20 arc*seconds*), the problem is that a typical all-sky reference catalog like Gaia is going to contain ~ 1 star.  So you will definitely need to create custom index files, using deep images of your own.  (Even a fairly deep large-area sky survey like the DESI Legacy Surveys won't contain enough stars.)  The build-astrometry-index program does this.

And to clarify, the code doesn't demand two quadrangle matches, it uses a single quadrangle (or triangle or quint, depending on the index file), but then does a "verification" step where it says "if this quadrangle match is correct, where else would I expect to find stars", omitting the quadrangle stars, and then demands that an odds-ratio test between a correct match and a coincidental match has large odds in favor of a correct match (like 10^9-to-1 odds).  You can turn down that odds ratio to potentially allow more positive results (true and false!).  But even for images like 10' x 10', every single quadrangle will match to a quadrangle on the sky.  (This reduces if you set --scale-low and --scale-high, or --ra,--dec,--radius).

cheers,

dustin

On Thu, Dec 8, 2022 at 8:15 PM Violet Felt <viole...@gmail.com> wrote:

Hi Dustin,

This is awesome! I'm a grad student at MIT trying to fit 20" x 20" frames, which often don't have eight stars in them. I understand the need for matching multiple quads when you have no pointing information to remove false positives, but is it possible to add an option to fit with just one quad when pointing information is available?

I'm willing to dig into the code and try and modify it myself if necessary, but I figured it wouldn't hurt to ask you first!

Thanks,

Violet

--
You received this message because you are subscribed to the Google Groups "astrometry" group.
To unsubscribe from this group and stop receiving emails from it, send an email to astrometry+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/astrometry/49612b46-b098-478d-88bd-74f4fd5f0966n%40googlegroups.com.