Author Topic: Background Extraction Challenge  (Read 516 times)

Offline midnightlightning

  • Newcomer
  • Posts: 19
    • View Profile
Background Extraction Challenge
« on: 2019 April 28 01:51:17 »
Hi,

I am really struggling to remove gradients in my B33 area RGB due to there being little or no natural background in the image. Can anyone suggest a way forward.
I have attached a stretched jpeg of the red channel, the individual RGB Masters are too big to post.

Thanks
Jon


Offline pfile

  • PTeam Member
  • PixInsight Jedi Grand Master
  • ********
  • Posts: 4535
    • View Profile
Re: Background Extraction Challenge
« Reply #1 on: 2019 April 28 09:43:44 »
you can upload your masters to google drive or dropbox (or similar) and post a link here.

in the meantime i wonder if all of your subs are good as the gradient you have does not seem like light pollution. did you blink thru your calibrated subs to see how they all look?

rob

Offline chris.bailey

  • PixInsight Addict
  • ***
  • Posts: 235
    • View Profile
Re: Background Extraction Challenge
« Reply #2 on: 2019 April 28 14:21:51 »
Indeed. Why do you think there are any gradients or brightness anomalies in that. Plenty of images out there show similar darkening in that area so you may be trying to DBE out real data.

Offline midnightlightning

  • Newcomer
  • Posts: 19
    • View Profile
Re: Background Extraction Challenge
« Reply #3 on: 2019 April 29 02:22:44 »
Hi,

Yes I used Blink and SFS prior to processing and dropped poor images, the rest looked fine.

The original Red image I posted was over stretched to emphasise the nebulosity, these images all have standard STF applied.

I think I see a gradient from bottom right to top left but stand to be corrected, the Ha data looks fine without background extraction.

I have since tried ABE which seems to do a better job than my DBE attempts at evening out th egradient but adds loads of noise so I am concerned I may be losing good data, also I am doing this before Channel Combine, perhaps worth trying after combining?

On a different topic, when running Deconvolution on my Ha it is introducing what look like hundreds of hot pixels – any idea why and how to stop it, there are no other artefacts and it otherwise looks good.
EDIT - I just fixed the Decon with 0.01 Global Bright :)

EDIT 2 - I would still be interested in views but I think I may have found a solution. I have Channel Combined RGB and then used DBE. I used a fully processed Ha image to identify areas of neutral sky, just a few small areas around the edges, and placed samples in the same place on the RGB - it seems to have worked without adding too much noise.

Thanks
Jon
« Last Edit: 2019 April 29 05:05:37 by midnightlightning »

Offline pfile

  • PTeam Member
  • PixInsight Jedi Grand Master
  • ********
  • Posts: 4535
    • View Profile
Re: Background Extraction Challenge
« Reply #4 on: 2019 April 29 08:04:23 »
your 2nd solution is good, it's a known technique (find sample points on a clean image, then apply to image needing extraction)

neither ABE or DBE can add noise to the image. but what happens is that since the DBE/ABE image is missing a (somewhat) constant offset, when you compute the STF for the DBE/ABE image, it is naturally more aggressive and reveals more of the noise that was already in the image. also, there's a certain amount of noise in the sky signal that was removed, but both DBE and ABE remove a smoothed version of the sky signal which leaves just the sky noise hanging around.

you can prove this to yourself by computing the STF for the pre-DBE'd image and saving it, then after DBE/ABE, apply that same STF to the new image and see what it looks like vs. re-doing STF on the image. you'll see that the image with the old STF does not seem as noisy as the post-DBE STF'd image.

rob

Offline midnightlightning

  • Newcomer
  • Posts: 19
    • View Profile
Re: Background Extraction Challenge
« Reply #5 on: 2019 April 29 14:39:09 »
Thanks Rob, I will try that :)