which method is better for denoising image with Gaussian noise?

i want to denoising image with Gaussian noise

Answer 1

[Image Analyst]

Statistically speaking, a simply local averaging filter would provide the maximum likelihood answer. However that will also blur true edges. There are other simple filters like median filter, Kuwahara filter (demo attached), as well as others for images that are considerably more sophisticated and better like BM3d, non-local means, K-SVD, K-LLD, UINTA, etc.

See this comprehensive overview by one of the leading denoising scientists: https://users.soe.ucsc.edu/~milanfar/publications/journal/ModernTour.pdf

[Mohammadreza]

Thanks a lot for your guide line.

[Image Analyst]

There is a new function added in R2018b: imnlmfilt().

It does non-local means denoising, which is generally regarded as one of the best image denoising filters out there.