MURE Denoise Observation

[jerryyyyy]

I have become a big fan of MURE as I have learned Chris Woodhouse's book that it is a good idea to clean noise up before you stretch your image (and noise).  It has been pretty easy for me with H-alpha but I have come to the cosmic realization that I should apply it to the RGB images individually also before they are combined and stretched.  Duh.

In any case, for my STT 8300M the DetectorGaussianNoise is about 30 for the H-alpha... if I double that to 60 it pretty much just works on the RGB images.... 30 is too little. 

My DetectorGain is about 0.40.

Your mileage may vary.

[mschuster]

Thank you.

Note that the noise parameters are detector specific and filter independent. Doing what you are doing is OK as long as you are happy with the results, but please check carefully for denoising artifacts and loss of weak signal SNR.

Also note that at least for sky limited wideband RGB, noise in the sky dominates detector Gaussian noise by definition, so the Gaussian noise parameter should have small impact on the denoised results, as image noise is primarily photon related and hence primarily detector gain dependent.

Denoising prior to processing makes the process of guessing what is noise and what is signal more accurate. Such denoising emulates integrating more frames before processing.

Regards,
Mike

[rdryfoos]

Thank you.

Note that the noise parameters are dectector specific and filter independent. Doing what you are doing is OK as long as you are happy with the results, but please check carefully for denoising artifacts and loss of weak signal SNR.

Also note that at least for sky limited wideband RGB, noise in the sky dominates detector Gaussian noise by definition, so the Gaussian noise parameter should have small impact on the denoised results, as image noise is primarily photon related and hence primarily detector gain dependent.

Denoising prior to processing makes the process of guessing what is noise and what is signal more accurate. Such denoising emulates integrating more frames before processing.

Regards,
Mike

   Hello--The literature indicates that to estimate the noise you use 2 darks (or biases) and get a number from the estimator script.  Since darks are filter independent, how is mure denoise filter dependent?

Thanks Rodd

[mschuster]

Hi Rodd, the script's detector noise parameters are not filter dependent. The filter reduces photon count, it looks to the script like a less exposed frame, i.e. lower SNR.

[mmirot]

DetectorGaussianNoise ?

Where can I find this script?

Max

[mschuster]

Use DarkBiasNoiseEstimator script to estimate detector Gaussian noise. Two biases or two darks needed.

Use FlatSNREstimator script to estimate detector gain. Two flats and one bias needed.