MATLAB: Would any one suggest some of best nonlinear filter out their for noise reduction in digital image processing (spatial/frequency domain, for which kind of noise doesnt matter)

Question

In general filtering of digital images there are linear and nonlinear filtering techniques in that i want to know some of best nonlinear filtering that is being used for the image reduction, for which type of noise doesnt not matter here,just want the names..

Answer 1

From what I've seen, the current state of the art is BM3D. At least that's what everyone was comparing their algorithms (usually unsuccessfully) to a few years ago. There may be better ones since.

Another good one is "non-local means." I think there are File Exchange versions of that.

There are others, like K-SVD, K-LLD, and UINTA. And new ones come up every year that try to topple BM3D. Certain algorithms may be better for certain types of noise or if you know certain things in advance about the noise (like its spectral distribution or other properties). I doubt there is one algorithm that is best for any and all situations.

Professor Peyman Milanfar is one of the leading gurus in the field of image denoising and he has written an overview that you should look at: https://users.soe.ucsc.edu/~milanfar/publications/journal/ModernTour.pdf