New PixInsight Tutorial on Noise Reduction with 4 Processes

[kayronjm]

Read here: http://www.lightvortexastronomy.com/tutorial-noise-reduction.html

This PixInsight tutorial covers the very important procedure of noise reduction, exploring one of PixInsight's core strengths. Four processes are covered in detail, namely - MultiscaleLinearTransform, MultiscaleMedianTransform, ACDNR and TGVDenoise. Specific instructions and applicable examples are given to guide the reader through the use of each process. Monochrome and colour images are both covered.

[UPDATE]: Changed tutorial to cover MultiscaleLinearTransform instead of ATrousWaveletTransform, since the former works identically using the Starlet transform algorithm and the latter is being phased out to obsolete by the PixInsight developers. Added brief discussion of each of the algorithms provided by both MultiscaleLinearTransform and MultiscaleMedianTransform. Added a fifth section to cover extraction of wavelet layers from an image to analyse noise at different scales.

[Juan Conejero]

Hi Kayron,

Please note that ATrousWaveletTransform is obsolete in all current versions of PixInsight. This process (along with the even older ATrousWaveletTransformV1) is only kept for compatibility with existing projects, icon files and scripts generated years ago. It will be moved to the Obsolete category in the next version (it belongs now to the Compatibility category), and eventually removed in a future version.

ATrousWaveletTransform has been replaced with MultiscaleLinearTransform, which implements two multiscale transforms (starlet transform, equivalent to the old à trous, and multiscale linear transform) and a linear mask feature identcal to its counterpart in MultiscaleMedianTransform.

Also note that ACDNR is also obsolete, although it still works quite well in many practical cases for deep-sky images. There are no plans to formally obsolete or remove it, though.

Finally, MultiscaleMedianTransform implements two different transforms: the 'pure' multiscale median transform and the hybrid median-wavelet transform. The advantages of the latter are summarized in the tooltip text available on the tool.

[kayronjm]

Hola Juan!

Thank you for your details. I appreciate them. I was aware that ATrousWaveletTransform would eventually become obsolete. In the interest of future-proofing, I will probably add MultiscaleLinearTransform or rather replace ATrousWaveletTransform with it and mention what you just said. Thank you again!

Best Regards,
Kayron

[Nocturnal]

Thank you! I found the MultiScaleLinearTransform NR tutorial really helpful! You explained really well what was going on and luckily even using the same parameters gave me very nice NR effects on a pretty noisy linear image I had.