Robust estimation approach for NL-means filter

Abstract

Edge preserved smoothing techniques have gained importance for the purpose of image denoising. A good edge preserving filter is given by NL-means filter than any other linear model based approaches. Since the weight function in NL-means filter is closely related to the error norm and influence function in robust estimation framework, this paper explores a refined approach of NL-means filter by using robust estimation function rather than the usual exponential function for its weight calculation. Here the filter output at each pixel is the weighted average of pixels in the surrounding neighborhoods using the chosen robust M-estimator function. Validations using various test images have been analyzed and the results were compared with the other known recent methods. There is a reason to believe that this refined algorithm has some interesting and notable points. © 2008 Springer Berlin Heidelberg.