On the random walks algorithms for image processing

Abstract

Random walks are time-reversible and finite special Markov chains. They are used to model different phenomena in mathematics and physics widely applied in many fields as economics, ecology, biology, computer science, etc. In this paper, we present a review of the use of random walks algorithm in images processing, called, random walks on graphs [7]. The basic concept of this algorithm remains in considering a given undirected graph G = ( V, E ), the random walk is a stochastic process that starts at a given vertex and move to a randomly selected neighbour, from which a similar process will be executed. The sequence of selected vertices defines a random walk on the graph. Random walks was used widely to solve several image processing problems, we cite here, image enhancement, image segmentation and image registration.