I have recently moved to using an ASI1600MM-Cool camera, a monochrome CMOS camera. Overall the camera is amazing, with exceptionally low read noise, low dark current, and performs wonderfully. I have noticed that dark frames are not catching all hot pixels, however. I think these hot pixels arise due to RTS, or Random Telegraph Signal, which seems to be an issue, albeit a small one, with the way CMOS sensors are read out. The thing about RTS is the pixels that may be affected are semi-random. The same pixel may appear "hot" in a few frames, but not all, and usually these pixels are averaged & clipped out of dark frames (even with a high sigma of 9 for Winsorized!)
So I've been using CosmeticCorrection with the auto-detect option, as well as manual mapping, to identify and correct the remaining hot pixels. It seems, however, that CosmeticCorrection is also missing hot pixels. It gets most, but there are still a few that stick around. I've run through the range of standard deviations with auto detect, and it does not seem to matter what setting I use, there are certain pixels that are just ignored. Here is an example...I've labeled the levels, 16-bit DN, in yellow for the missed hot pixel as well as it's neighbors. My sigma was 5.5, which is more than adequate to capture this particular hot pixel (as many others with similar deviations were corected):
I understand that generally speaking, these hot pixels can and should be caught and corrected by dithering and using a clipping algorithm in ImageIntegration. However, I have noticed that hot pixels like this can act as loci for artifacts when running ImageRegistration:
I can fiddle with ImageRegistration and eventually get it to not generate artifacts...however, that is a VASTLY more complex and lengthy process than simply running CosmeticCorrection on the frames before registration to just eliminate the source of the problem. I am very curious why CC is not properly identifying and rejecting these pixels, which are almost always single, individual pixels (but rarely they may form in a small group), and easily identifiable as something other than a star.
Can this issue be fixed?
