BAYESIAN DECISION-FEEDBACK FOR SEGMENTATION OF BINARY IMAGES

We present real-time algorithms for the segmentation of binary images modeled by Markov mesh random fields (MMRF's) and corrupted by indepen dent noise. The goal is to find a recursive algorithm to compute the m aximum a posteriori (MAP) estimate of each pixel of the scene using a fixed lookahead of D rows and D columns of the observations. First, th is MAP fixed-lag estimation problem is set up and the corresponding op timal recursive (but computationally complex) estimator is derived. Th en, both hard and soft (conditional) decision feedbacks are introduced at appropriate stages of the optimal estimator to reduce the complexi ty. The algorithm is applied to several synthetic and real images. The results demonstrate the viability of the algorithm both complexity-wi se and performance-wise, and show its subjective relevance to the imag e segmentation problem.