Roundness filter

Discussions of Electronically Assisted Astronomy using the Live Stacking feature.
Post Reply
Yourjones
Posts: 11
Joined: Mon Jun 21, 2021 9:14 am

Roundness filter

#1

Post by Yourjones »

Hello

I understand the FWHM filter is a powerful tool to filter out frames based on quality of star shapes/sizes. I wonder if this algorithm also take star roundness into consideration. Or is there the possibility of adding such a filter in the future updates? :lol:

Thank you ver much!

Lu
User avatar
admin
Site Admin
Posts: 13173
Joined: Sat Feb 11, 2017 3:52 pm
Location: Vale of the White Horse, UK
Contact:

Re: Roundness filter

#2

Post by admin »

Hi,

currently the algorithm looks at the roundness (but only to discard stars that are very elongated - these would potentially be other things like double stars and would give incorrect FWHM readings).

The same idea of a roundness filter was suggested in a different thread yesterday. It would be interesting if you found some sample frames - a good one with round stars and a bad one with elongated stars. If you could share them here then I can investigate how well the code might be able to tell between them.

Robin
Yourjones
Posts: 11
Joined: Mon Jun 21, 2021 9:14 am

Re: Roundness filter

#3

Post by Yourjones »

Hi Robin

Thanks for your interest in the expected roundness filter. Unfortunately, I don't have really round stars that I'm satisfied with, but I've got plenty of those with stars that are not round. Should I need, I'd be happy to collect some "round" vs "not round" stars when I run into my next imaging session. Thanks again!

Lu

Crop 1: REALLY, REALLY BAD
not round 1.jpg
not round 1.jpg (981.28 KiB) Viewed 1865 times
Crop 2: BAD
not round 2.jpg
not round 2.jpg (978.93 KiB) Viewed 1865 times
Crop 3: EQUALLY BAD
not round 3.jpg
not round 3.jpg (971.26 KiB) Viewed 1865 times
User avatar
admin
Site Admin
Posts: 13173
Joined: Sat Feb 11, 2017 3:52 pm
Location: Vale of the White Horse, UK
Contact:

Re: Roundness filter

#4

Post by admin »

Hi,

just to follow up on this, I have taken some time to do testing both of these images and some sample images that I have deliberately blurred in one direction to simulate poor tracking/wind/mount shake.

Using a fairly standard measure of roundness (circularity, defined as 4 * pi * area / perimiter squared), I have not seen much sensitivity of the circularity measurement to how 'non round' the stars are. In fact in some cases smearing the image diagonally by four pixels increases the measured circularity :(

In contrast to this, FWHM measurements are sensitive to non-round stars being stretched out, with a clear increase in FWHM even just by adding 10% of the brightness of the image with a 1 pixel shift.

So, for now I'm going to put any more work on a roundness filter on hold as it doesn't seem to bring any benefit over the current FWHM filter. If we can collect some samples that show the opposite then I can consider having another look.

cheers,

Robin
Post Reply