TREE APPROXIMATIONS TO MARKOV RANDOM-FIELDS

Methods for approximately computing the marginal probability mass func tions and means of a Markov random field (MRF) by approximating the la ttice by a tree are described. Applied to the a posteriori MRP these m ethods solve Bayesian spatial pattern classification and image restora tion problems. The methods are described, several theoretical results concerning fixed-point problems are proven, and four numerical example s are presented, including comparison with optimal estimators and the Iterated Conditional Mode estimator and including two agricultural opt ical remote sensing problems.