Author Topic: Linear fit clipping and Winsorized sigma clipping  (Read 6947 times)

Offline Jules

  • PixInsight Guru
  • ****
  • Posts: 513
Linear fit clipping and Winsorized sigma clipping
« on: 2012 March 20 14:51:10 »
Hi

Please have a look at the two attached images. It is the same image made up from 17 Ha sub frames of 10 and 20 minutes. I have dark/bias/ flat fielded the subs then star aligned.

The problem I am seeing is white pixels (see image) when using Linear Fit Clipping. When using Winsorized Sigma Clipping the white pixels do not appear.

These groups of white pixels appear all over the image.

My dark / bias frames work very well, I have been using both rejection algorithms on default.

Regards

Julian


Offline georg.viehoever

  • PTeam Member
  • PixInsight Jedi Master
  • ******
  • Posts: 2132
Re: Linear fit clipping and Winsorized sigma clipping
« Reply #1 on: 2012 March 21 01:08:14 »
Have a look at your raw/calibrated data to see if there is something special in these locations... Also, you may need to use different rejection values for linear fit clipping.

Georg
Georg (6 inch Newton, unmodified Canon EOS40D+80D, unguided EQ5 mount)

Offline Cleon_Wells

  • PixInsight Addict
  • ***
  • Posts: 225
Re: Linear fit clipping and Winsorized sigma clipping
« Reply #2 on: 2012 March 21 06:00:38 »
Julian, have you read through this documentation on ImageIntegration and its many pixel rejection settings?

http://pixinsight.com/doc/tools/ImageIntegration/ImageIntegration.html

I don't try to understand the math,  but I do get a better idea of what's going on with PixelRejection by looking at the graphs. The Documentation section

http://pixinsight.com/doc/tools/

is a great place to gain a foothold on the PI mountain.   :)

Cleon
Cleon - GSO 10"RC/Canon T1i-Hap Mod, 100mmF6/2Ucam/MG, EQG/EQmod

Offline Jules

  • PixInsight Guru
  • ****
  • Posts: 513
Re: Linear fit clipping and Winsorized sigma clipping
« Reply #3 on: 2012 March 21 06:44:51 »
Hi Georg and Cleon

I thought I new Image Integration pretty well especially linear fit clipping. For my imaging I try to get at least 15 subs then experiment with sigma high/low to optimise SNR.

Looking at the subs there is nothing specific at these locations?

Regards

Julian

Offline zvrastil

  • PixInsight Addict
  • ***
  • Posts: 179
    • Astrophotography
Re: Linear fit clipping and Winsorized sigma clipping
« Reply #4 on: 2012 March 21 07:35:13 »
These white points looks similar to VNG-debayered hotpixels. I suppose your images are grayscale and there's no debayering involved. Have you checked all of your subs? I'd doublecheck if there are hotpixels in subs in those areas (maybe just in the one sub?). I can't imagine how'd image integration (no matter which algorithm) somehow invent such random pixels.

regards, Zbynek

Offline Jules

  • PixInsight Guru
  • ****
  • Posts: 513
Re: Linear fit clipping and Winsorized sigma clipping
« Reply #5 on: 2012 March 21 13:00:03 »
Hi Zbynek

I have checked all the subs in this image, out of the 16 there are 3 with the randomly placed white pixels.

Regards

Julian