Textured image segmentation using multiresolution Markov random fields

A new stochastic image model, based on a multiresolution representation, is presented and used in the Bayesian segmentation of textured images. Results on a variety of images are used to demonstrate its effectiveness.