A Nonlocal Poisson Denoising Algorithm Based on Stochastic Distances

Abstract

In this letter, a new version of the Nonlocal-Means (NLM) algorithm based on stochastic distances is proposed for Poisson denoising. NLM estimates a noise-free pixel as a weighted average of image pixels, where each pixel is weighted according to the similarity between image patches. In this work, stochastic distances are used as a new similarity measure. We explored the use of four stochastic distances for which closed-form solutions were found for Poisson distribution. This approach was demonstrated to be competitive with related state-of-the-art methods.