Variational Bayes Approach For Tomographic Reconstruction

In this paper, we apply the Bayesian inference method in a tomographic reconstruction problem. For this purpose, we propose a Gauss-Markov field with Potts region label model for the images. Most of model parameters are unknown and we wish to estimate them jointly with the object of interest. Using the variational Bayes framework, the joint posterior law is approximated by a product of marginal laws whose shaping parameter equations are derived. An application to tomographic reconstruction is presented with discussion of convergence and quality of this estimation.