A Bayesian approach for texture images classification and retrieval

Texture analysis plays an important role in many image processing and computer vision tasks, ranging from natural to medical imaging and content-based image retrieval. In this paper, we present an efficient Bayesian algorithm for texture image classification and retrieval, based on Reversible Jump Markov Chain Monte Carlo (RJMCMC) and general Beta mixture models. Our work is motivated by the fact that textured images are generally described by non-Gaussian characteristics which cannot be realistically modeled using rigid distributions. Beta mixtures are able to fit any unknown distributional shape and then can be considered as a useful and flexible solution for the problem of modeling non-Gaussian features present in texture images. In theory, it is well-known that full Bayesian approaches, to handle the mixture estimation and selection problems, are fully optimal. We applied then a fully Bayesian, RJMCMC, technique which simultaneously allows cluster assignments, parameters estimation, and the selection of the optimal number of clusters. Experimental results involving a challenging texture images data set are presented and discussed to show the merits of the proposed work.