Author Topic: Signal to noise ratio  (Read 3255 times)

Offline ssavignac

  • Newcomer
  • Posts: 13
Signal to noise ratio
« on: 2013 October 18 13:38:29 »
Hello,

How can I evaluate the signal to noise ratio after staking.
Is it a good way to know if the image worth to work on further.

Samy

Offline pfile

  • PTeam Member
  • PixInsight Jedi Grand Master
  • ********
  • Posts: 4729
Re: Signal to noise ratio
« Reply #1 on: 2013 October 18 14:07:02 »
ImageIntegration computes the SNR deltas after stacking; the results are in the console window after it's done.

also you can always use the NoiseEvaluation script

rob

Offline ssavignac

  • Newcomer
  • Posts: 13
Re: Signal to noise ratio
« Reply #2 on: 2013 October 19 05:50:32 »
Thank you,

What is a good number we looking for in the script.

the number I saw are quit different from then console windows.

I Have between 40% and 52%.

Samy

Offline mschuster

  • PTeam Member
  • PixInsight Jedi
  • *****
  • Posts: 1087
Re: Signal to noise ratio
« Reply #3 on: 2013 October 19 07:36:03 »
Samy,

Multiply the first number the script prints (i.e. the number in scientific notation) by 65535 to get an rms noise estimate in ADU. Ignore the other numbers.

Then open the Statistics process and read the AvgDev value. Choose the 16-bit range so the value is also in ADU.

Then divide these two values, to get an SNR estimate.

This estimate has some limitations which affect its validity. Only compare the SNR estimates of subframes or integrations of exactly the same target (small dithering and alignment differences are OK). When comparing subframes, wildly different gradients will invalidate the comparison. There are other issues also.

Various processes estimate SNR differently and possibly more robustly. This is only one way to do it.

Mike
« Last Edit: 2013 October 19 08:16:51 by mschuster »