Mean field decomposition of a posteriori probability for MRF-based image segmentation: Unsupervised multispectral textured image segmentation

This paper proposes a Markov random field (MRF) model-based method for unsu pervised segmentation of multispectral images consisting of multiple textur es. To model such textured images, a hierarchical MRF is used with two laye rs, the first layer representing an unobservable region image and the secon d layer representing multiple textures which cover each region. This method uses the Expectation and Maximization (EM) method for model parameter esti mation, where in order to overcome the well-noticed computational problem i n the expectation step, we approximate the Baum function using mean-field-b ased decomposition of a posteriori probability. Given provisionally estimat ed parameters at each iteration in the EM method, a provisional segmentatio n is carried out using local a posteriori probability (LAP) of each pixel's region label, which is derived by mean-field-based decomposition of a post eriori probability of the whole region image. Experiments show that the use of LAPs is essential to perform a good image segmentation.