Does BackgroundNeutralization subtract or divide?

[M Covington]

Does BackgroundNeutralization divide R, G, and B by appropriate factors to correct a color cast, or does it subtract constants from them?

I would have expected the former, but when I try to replicate it with AutomaticBackgroundExtractor, I find that if I choose Subtraction, I get the same thing as with BackgroundNeutralization, but if I choose Division, I get more complete correction of the color cast (note white rather than cream-colored stars, and both bluish and yellowish areas of the galaxy distinguishable).

For color cast removal I set the function degree in AutomaticBackgroundExtractor to 0 so it won't try to find gradients, but as far as color is concerned, results are the same if the degree is set to a more normal value such as 2 to 4.

 

[pfile]

based on line 255 of the following file:

PCL/src/modules/processes/ColorCalibration/BackgroundNeutralizationInstance.cpp at master · PixInsight/PCL

PixInsight Class Library. Contribute to PixInsight/PCL development by creating an account on GitHub.

github.com

it looks like the actual background neutralization operation is addition (subtraction.) i don't see any division happening there, so your experiments match with how i read the code.

but does it matter? BN is not the entire color calibration flow but instead a necessary prerequisite to color calibration.

rob

 

[M Covington]

Thanks. That is very helpful. I had forgotten that the code for this was available.

 

[Juan Conejero]

The GitHub mirror of our open-source repository is obsolete and no longer used. It will be removed soon.

All of our source code repositories are on GitLab since February 2019:

PixInsight · GitLab

Advanced Image Processing Platform https://pixinsight.com/

gitlab.com

 

[pfile]

sorry- i know that but i just quickly googled for the code and the old repo was the top hit.

rob