Source continuity and boundary discontinuity considerations in Bayesian image processing

This paper extends the Bayesian image processing (BIP) formalism by considering the effect of simple source continuity and boundary discontinuity and a priori information in estimating an optimal source distribution from observed data. The a priori source information is formulated in terms of probability density functions of source element strengths and spatial correlations. The estimation is carried out iteratively by a BIP algorithm derived by applying the expectation maximization technique to the a priori source probability density functions and assuming the data obey Poisson statistics. The suppression of boundary oscillations and enhancement of overall image are demonstrated for computer generated ideal and Poisson randomized data.