Textured image segmentation via neural network probabilistic modeling

It has been shown that a trained backpropagation neural network (BPNN) classifier produces outputs which can be interpreted as estimates of Bayesian a posteriori probabilities. Based on this interpretation, the BPNN approach for the estimation of the local conditional distributions of textured images, which are commonly represented by a Markov random field (MRF) formulation, is presented. The proposed BPNN approach overcomes many of the difficulties encountered when using an MRF formulation. The approach does not require the trial-and-error choice of clique functions or the subsequent unreliable estimation of clique parameters. Simulations show that the images synthesized using BPNN modeling produce desired textures more consistently than MRF-based methods. The application of the proposed BPNN approach to synthesis and segmentation of real world textures is presented