Efficient image understanding based on the Markov random fieldmodel and error backpropagation network

摘要

Image labeling is a process of recognizing each segmented region,properly exploiting the properties of the regions and the spatialrelationships between regions. In some sense, image labeling is anoptimization process of indexing regions using the constraints as to thescene knowledge. This paper further investigates a method of efficientlylabeling images using the Markov random field (MRF). MRF model isdefined on the region adjacency graph and the labeling is then optimallydetermined using simulated annealing. The MRF model parameters areautomatically estimated using the error backpropagation network. Theauthors analyze the proposed method through experiments using the realnatural scene images