Deconvolution approach and poor seeing conditions

Brian Valente

Hey Adam & all,

i'm wondering if deconvolving the LUM channel prior to integration with RGB may not be the best approach for my images?  i image in bortle 11, deep urban los angeles. In addition to low SNR, the pollution also results in less than sharp images.

I have been following the Fundamentals processing steps for my OSC and mono cameras, including processing the LUM channel different than process for RGB. 

however, i've found my red channel is much more bloated than any other channel (both in OSC and mono), so when I combine the lum and RGB images, the stars end up looking like they have a red halo around them. 

What do you think about deconvolving the whole thing after lum+rgb integration, instead of just lum prior to integrating?

Brian

Brian Valente

Here's a link to fully processed (but not yet stretched) Lum and RGB images:

Lum  https://www.dropbox.com/s/111ufr23w4mqo1q/M56 Lum unstretched.xisf?dl=0

RGB - https://www.dropbox.com/s/sypgv050y5dbjtg/M56 RGB unstretched.xisf?dl=0 

BlockHead

Hi Brian,

Hmm... let me set aside the OSC part of your issue above...

Concerning a "normal" RGB image created from mono R,G,B data- large variations in seeing can certainly cause an issue like you describe. I do not think deconvolution of Luminance is somehow making this worse on a per channel basis. What could be happening is that a sharpened luminance blended with soft RGB data can certainly make some stellar halos- and if you have a color bias in the star size (with color fringing) this will just be more apparent- but not a cause per se.

In this situation in the past I have applied an erosion filter (Morphological minimum) to the offending color channel. You could select only the stars first so that nothing else is affected. A very small amount will take care of the offending color fringe. In this way, by operating on only the chrominance... you will not be affecting the luminance image at all.

Now with regards to OSC... it is pretty hard to get a color mis-match in stellar profile since the data is all acquired simultaneously. If you are seeing red halos (or whatever color) in your final LRGB... they were already there in the RGB image to being with...but perhaps made easier to see. But I would be surprised to see this with OSC data. 

I did look at the images.. they look good... I see the redness around stars in the RGB... though it isn't crazy. In fact...with your permission...I will make a quick video to show the "technique." FWIW

-the Blockhead 

Brian Valente

Thanks Adam - i appreciate the input i will look into the erosion filter

if you want to do a video you are welcome to use/distribute my data 

Brian