Is there a way to filter out frames with tracking errors

Discussions of Electronically Assisted Astronomy using the Live Stacking feature.
Post Reply
jeff2011
Posts: 12
Joined: Sun Jun 10, 2018 4:10 pm

Is there a way to filter out frames with tracking errors

#1

Post by jeff2011 »

Our club is doing EAA on an old Equatorial fork mount that holds a large 18 inch newt. It does well for the most part but occasionally has a tracking glitch that results in a frame with streaked stars. The FWHM filter does not filter these out as the stars prior to the glitch are bright enough to align on and their FWHM are less than that of a frame without the streak. Sigma rejection helps but I was wondering if there was a way to reject these frames entirely.
User avatar
admin
Site Admin
Posts: 13173
Joined: Sat Feb 11, 2017 3:52 pm
Location: Vale of the White Horse, UK
Contact:

Re: Is there a way to filter out frames with tracking errors

#2

Post by admin »

Hi,

currently there's nothing beyond the FWHM and sigma clipping (unless perhaps the star brightness threshold might help - maybe the streaked stars show up as dimmer?).

If you can save a few raw frames - ones with and without the streaking - and share them with me, they would make a good test for future improvements in this area. I have been thinking about looking for streaks in the image before stacking (although more from the point of view of looking for and removing satellite trails), but it may be that some sort of streak detection could be used for both things.

cheers,

Robin
jeff2011
Posts: 12
Joined: Sun Jun 10, 2018 4:10 pm

Re: Is there a way to filter out frames with tracking errors

#3

Post by jeff2011 »

Thanks Robin that would be a useful feature. We save our frames for replay at outreach on cloudy nights. I am sure we have some saved with streaking stars. I will gather some sample frames and get them to you next time I am out at my club's observatory.
User avatar
admin
Site Admin
Posts: 13173
Joined: Sat Feb 11, 2017 3:52 pm
Location: Vale of the White Horse, UK
Contact:

Re: Is there a way to filter out frames with tracking errors

#4

Post by admin »

Hi Jeff,

that would be great - thanks :) I will obviously create some frames like this myself too, but getting multiple sources helps ensure that I don't just write code that can fix the errors that I have deliberately introduced and nothing else!

cheers,

Robin
jeff2011
Posts: 12
Joined: Sun Jun 10, 2018 4:10 pm

Re: Is there a way to filter out frames with tracking errors

#5

Post by jeff2011 »

Sorry it took so long Robin. Below is a link to an example. The focus is poor but it illustrates the problem. For the frames with bad tracking it will calculate a FWHM smaller than the other frames and will stack that bad ones that it finds enough stars. To live stack you will need to increase digital gain and increase the allowed star size and set the minimum number of stars to stack to 10.

https://1drv.ms/u/s!AkmD-T87yO8QpnMkjR0 ... Q?e=Jeg0vy
User avatar
admin
Site Admin
Posts: 13173
Joined: Sat Feb 11, 2017 3:52 pm
Location: Vale of the White Horse, UK
Contact:

Re: Is there a way to filter out frames with tracking errors

#6

Post by admin »

Hi Jeff,

thanks for taking the time to collect those frames - I can see that frame 310 is badly trailed by comparison to the others - I will have a play at running various tests on the star detection on the trailed frames vs the good ones to see if the trailing can be detected.

cheers,

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

Re: Is there a way to filter out frames with tracking errors

#7

Post by admin »

Ah, OK, at least I understand now why the streaked frames are lower FWHM...

Most of the problem frames are not a uniform trailing streak for the star, instead they are breaking the star up into 3 or 4 positions as the tracking glitches. It's as though the mount tracks for a few seconds, glitches, tracks a few more seconds, repeat. You can see this below
Capture.JPG
Capture.JPG (162.63 KiB) Viewed 1393 times
What you will also notice is that the star detection is picking up on the individual bright spots in the star trail, rather than the trail as a whole. These seem to have a tighter pattern than the stars in the good frames (perhaps because you have some general tracking wobble in the background, so even the good frames are spread out somewhat).

Looking at the frames, I don't think it's realistic to expect to be able to get code that can reliably spot the difference and 'fix' this data in software - there is definitely something odd going on in your mount tracking (mechanically) - perhaps the movement sticks momentarily and then frees up again. I think that's the place to try to attack this problem. Note that since your frames seem to be 10s exposures and most of the problem frames have 3 or 4 bright spots in each star trail, it looks like the mechanical glitch is occuring every 2-3 seconds on average.

Sorry not to be able to help more :(

cheers,

Robin
jeff2011
Posts: 12
Joined: Sun Jun 10, 2018 4:10 pm

Re: Is there a way to filter out frames with tracking errors

#8

Post by jeff2011 »

Thanks for taking your time to look into this Robin.
User avatar
admin
Site Admin
Posts: 13173
Joined: Sat Feb 11, 2017 3:52 pm
Location: Vale of the White Horse, UK
Contact:

Re: Is there a way to filter out frames with tracking errors

#9

Post by admin »

Actually, I thought about this a bit more...

it may be possible to get some filtering to work on this using the existing functionality in SharpCap

1) Have a play with the brightness filter in live stacking - this is designed to reject frames if the average star brightness drops (in particular in case of thin cloud appearing intermittently), but it might pick up on the fact that the stars being found in the bad frames are dimmer than in the good quality frames

2) Another possibility would be to reduce the sensitivity of the star detection deliberately until it only just works for the good frames. Then maybe the bad frames would not even detect enough stars to align, so they would also get ignored by the live stacking.

hope this helps,

Robin
Post Reply