DISTRIBUTED PROPAGATION OF A-PRIORI CONSTRAINTS IN A BAYESIAN NETWORKOF MARKOV RANDOM-FIELDS

In this paper, Bayesian networks of Markov-random fields (BN-MRFs) are proposed as a technique for representing and applying a-priori knowle dge at different abstraction levels inside a distributed image process ing framework. It is shown that this approach, thanks to the common pr obabilistic basis of the two techniques, is able to combine in a natur al way causal inference properties at different abstraction levels as provided by Bayesian networks with optimisation criteria usually appli ed to find the best configuration for an MRF. Examples of two-level BN -MRFs are given, where each node uses a coupled Markov random field wh ich has to solve a coupled restoration and segmentation problem. Exper iments are concerned with expert-driven registered segmentation and tr acking of regions from image sequences.