Gibbs Random Image Models and Sampling

Abstract

I. INTRODUCTION Markov-type two-dimensional (2-D) random processes (or random fields) have been widely applied to many aspects of practical image processing such as image modeling [1-6], image restoration [7-13], image segmenta­ tion [14-20], and texture analysis and synthesis [21-25]. An excellent review by Derin and Kelly [26] discusses many Markov-type 2-D random field models and their interrelations applied to image processing.