Author Topic: Help With Noise Estimation  (Read 1422 times)

Offline GeneralT001

  • Member
  • *
  • Posts: 71
    • View Profile
Help With Noise Estimation
« on: 2016 August 21 12:37:14 »
Hi,

I am pretty new to Astro Imaging and to Pixinsight. I was hoping to get the voice of experience about noise in my images. Below is more or less a typical stack of 200 frames @ 30 sec each. The camera used was a OSC ASI 1600 cooled. The frames have been calibrated with flats/bias and darks. I have used Blink and the Sub Frame Selector to cull the frames as best I could. They were then registered and integrated.

Is the amount of noise, in the image, fairly typical of what to expect given a relatively short integration (1hr40min)? I have zoomed in a fair bit. I do have some light pollution (think I am in a yellow zone). I have tried dithering as well and the results are pretty much the same.

Can this noise be removed successfully? If so, is it best done in a linear state (MultiScaleTransform?) or in a non-linear state (ACDNR?). Or done in both states? Which is the best method (work flow) to attack this noise?

Would a mono camera reduce the amount of noise the images seem to have?

Appreciate any thoughts about this.


Many Thanks


Offline Alejandro Tombolini

  • PTeam Member
  • PixInsight Jedi
  • *****
  • Posts: 1267
    • View Profile
    • Próxima Sur
Re: Help With Noise Estimation
« Reply #1 on: 2016 August 21 15:14:23 »
Hi, i am looking with the phone but it seems for the kind of pattern that this processing example may be useful.
Saludos Alejandro

Offline GeneralT001

  • Member
  • *
  • Posts: 71
    • View Profile
Re: Help With Noise Estimation
« Reply #2 on: 2016 August 21 22:25:08 »
Hi Alejandro,

Thank you for the reply. I had a look at the link and tried the following, based on it:

My Darks were shot at the same time exposure and temp as the lights (30 sec and -15c). I was hoping that perhaps it was the reference frame that may have been the issue as I sometimes forget to note which has been weighted the highest in the Sub Frame Selector. But I reran the integration and positively confirmed that I was using the highest rated frame, but it made no difference, those streaks still showed up. I also ran an integration without the Sub Frame Selector at all but again, the same streaky result.

The streaks shows up in the Rejection Low/High frames of the Image Integration (see pic below). Not sure what else to try, will redo the Calibration frames and retry dithering to see if this will make a difference.


Thanks


Offline gvanhau

  • PixInsight Old Hand
  • ****
  • Posts: 345
    • View Profile
Re: Help With Noise Estimation
« Reply #3 on: 2016 August 22 06:33:03 »
Hello
This pattern seems to me a little drift between images.  if so, Mono CCD's will show the same streaks, and the best method to avoid it is dithering.

In order to remove the noise, you could try the FFT transform, remove the dominant pattern and then inverse FFT transform (simmilar to processing comet images).

Regards
Geert
« Last Edit: 2016 August 22 06:43:45 by gvanhau »
Geert Vanhauwaert

Offline Alejandro Tombolini

  • PTeam Member
  • PixInsight Jedi
  • *****
  • Posts: 1267
    • View Profile
    • Próxima Sur
Re: Help With Noise Estimation
« Reply #4 on: 2016 August 22 17:55:36 »
Hi, those hot pixels in the rejection map are on the calibrated image, which means that it was not correctly calibrated. Check your _c_r_d image and you should see the hot pixels moving on the image.
Try only one image in ImageCalibration tool with your MasterBias, MasterDark and MasterFlat, and check if the calibration works.
Saludos, Alejandro.