EDGE-PRESERVING TOMOGRAPHIC RECONSTRUCTION FROM GAUSSIAN DATA USING AGIBBS PRIOR AND A GENERALIZED EXPECTATION-MAXIMIZATION ALGORITHM

The problem of edge-preserving tomographic reconstruction from Gaussia n data is considered. The problem is formulated within a Bayesian fram ework, where the image is modeled as a pair of Markov Random Fields: a continuous-valued intensity process and a binary line process. The a priori information considered here enforces constraints both on the lo cal regularity of the image and on the line configurations. The soluti on, defined as the maximizer of the posterior probability, is obtained using a Generalized Expectation-Maximization (GEM) algorithm, in whic h both the intensity and the line processes are iteratively updated. T he simulation results show that introducing suitable priors on the lin e configurations improves the quality of the reconstructed images, and is particularly useful when the data record is small. The relationshi ps with other approaches for managing discontinuities are outlined. A comparison between the GEM algorithm and an algorithm based on mixed-a nnealing is made on the basis of computer simulations. (C) 1995 John W iley & Sons, Inc.