PageRank image denoising

Abstract

We present a novel probabilistic algorithm for image noise removal. The algorithm is inspired by the Google PageRank algorithm for ranking hypertextual world wide web documents and based upon considering the topological structure of the photometric similarity between image pixels. We provide computationally efficient strategies for obtaining a solution using the conjugate gradient algorithm. Comparisons with other state-of-art denoising filters, namely the total variation minimising filter and the bilateral filter, are made. © 2010 Springer-Verlag.