Noise Reduction

This documentation section focuses on processes specific to noise reduction. In PixInsight LE 1.0, those processes are SGBNR and SCNR.

PixInsight LE includes an improved implementation of SGBNR (Selective Gaussian Blur Noise Reduction), an efficient algorithm for reduction of noise at medium and large dimensional scales. This continues the tradition started in 2002 by the Pleiades Astrophoto Team with their popular SGBNR standalone application.

The second process specific to noise reduction implements the SCNR (Subtractive Chromatic Noise Reduction) algorithm. This is a technique developed to remove noise in the green channels of color deep-sky astrophotos.

But noise reduction is much more thoroughly covered by PixInsight LE. A powerful implementation of the à trous discrete wavelet transform algorithm, namely the ATrousWaveletTransform process, can be efficiently applied for noise reduction at small dimensional scales (high-frequency noise). With the implemented user interfaces, wavelets processing can be applied to perform both noise reduction and detail enhancement tasks at the same time.

Finally, optimized implementations of the median and erosion/dilation morphological filters are also very useful for specific noise reduction tasks, such as impulse noise removal.


SGBNR (Selective Gaussian Blur Noise Reduction)

SCNR (Subtractive Chromatic Noise Reduction)

Worked Noise Reduction Example: Wavelets + SGBNR