prusling le décembre 04, 2019, 09:33:59 09
Hi, I am having difficulty processing a set of narrowband data in that I'm finding it hard to remove hot pixels. First to say that I have quite a small number of subs with each filter - 6 x 10 mins - but realistically with the weather having been so bad getting a 3-hour session of clear sky is in itself something of a miracle!

I guess that some of my problems may be caused by the fact that I dither, but it's the cosmetic correction stage that seems not to find the hot pixels.

For each filter I'm using the following script generated from sirilic:

  preprocess images -dark=../MASTER/
  seqfind_cosme pp_images 3 3
  register cc_pp_images
  stack r_cc_pp_images rej 3 3 -nonorm  -out=../Ha

and it then registers the channels with each other before post-processing.

From a blink test of each channel, in one case the hot pixels have been removed but in the other two they remain.  I have adjusted the sigma values up/down but no change other than the channel that has the hot pixels removed sometimes changes. It also seems to change each run - maybe with a change in the reference frame? - the last run I did removed none of the hot pixels from any channel.

Please can you advise how best to process a limited number of dithered subs or whether dithering is not recommended at all.

Lastly, where the hot pixels are removed they are replaced by dark values - is it possible to have an option where they are replaced by the median of the surrounding pixels?

lock042 le décembre 04, 2019, 12:02:57 12
You need to update the threshold of detection. Here it is 3.
Also hot pixels are always replaced by median value. If you see black pixel it is because hot pixel has not been detected and subtracted by the one of the dark frame.

prusling le décembre 04, 2019, 01:50:28 13
OK thanks, what value for sigma would you recommend - higher, lower?

lock042 le décembre 04, 2019, 02:08:05 14
I don't know. You need to try.