›

Processing for correcting jail bar vertical FPN

Daniel Judge
in PixInsight
I have been going through the Fundamentals course and have an issue with FPN that I'm not sure how to attack.
I have an older KAI-11000 and was wondering if there was a method to further correct this that would work in pixinsight.
I can remove some of the FPN using LPS, but find that to finalise most removal I have to use the Astronomy Tools actions in Photoshop, "Vertical Banding Noise Reduction".  This does a good job in final clean up, but I would like to try and solve this in pixinsight if possible.
I also have quite an aggresive FPN that is noticable in a quater of the sensor plane that is quite hard to remove. This did get removed quite well in my luminance data, which I think was due to better illumination across the FOV.
I believe I might need to up the dithering to 30pixels from 20pixels to remove most of this further?
I was wondering if someone could provide further advice if further dithering would work best, or if someone might be up to the challenge of processing my images to see if a clean image can be obtained using LPS or some other 'magic'
Can provide dropbox link to all files if required.
Thanks
Below you can see my masters and integration times, using 20pixel random dithering during image acquisition.


Comments

  • BlockHeadBlockHead
    Once you are dithering to an spatial extent that is larger than the artifact... I doubt a larger dither will help much.

    That is some serious banding. I have not seen something this dramatic before. Yuck.

    I wonder if there is a way to create a "flat" (model the banding) in order to remove it.

    I don't have any good ideas that are better than what you are doing.
    You might need to look into exactly what PHotoshop actions are doing and reproduce them in PI using a combination of processes/PixelMath.

    -the Blockhead

  • Daniel Judge
    Yes, yuck indeed.

    I can see what the actions are doing, I will need to work out how to apply similar in pixinsight where possible.
    1.PNG
    316 x 1216 - 48K
    2.PNG
    303 x 220 - 7K
  • BlockHeadBlockHead
    ah!
    It is using High Pass Filter as a way to capture the banding. Then it uses it as a mask (and presumably corrects).That is my guess. If I am right, you can definitely do this in PI. Instead of a HighPass filter... you might use MLT to create capture the banding the create a mask. The hard part of the adjustment part... but High Pass FIlter has another property in that the color of the result itself has information. (captured structures are brighter than middle gray). So.. I am not certain exactly how it is working. 

    -the Blockhead
  • Daniel Judge
    After spending some time going through Fundamentals I've finally become somewhat more comfortable with processing my images.

    Trying to fix this slight issue in my images, I found that pixel math can assist in removing some pepper throughout the image, and remove those awful lines. I'm not sure if this process might be too aggresive in processing out detail. We'll see as we go.

    I first linear stretch the image to see the K value where this noise is present after WBPP. Then apply some pixelmath, to replace that noise value with a random value that is present in the avg surrounding background noise.

    :)




    before1.png
    2048 x 1374 - 1M
    after1.png
    2045 x 1369 - 1M
    before2.png
    2037 x 1367 - 1M
    after2.png
    2044 x 1366 - 1008K
    before3.png
    2051 x 1368 - 3M
    after3.png
    2039 x 1367 - 3M
    secretsauce.png
    770 x 249 - 12K
Sign In or Register to comment.