Bayesian-statistician approach in image restoration by means of solvable Markov random field model

摘要

We give the exact expression of the marginal likelihood in a solvable Markov random field model which have a form of Gibbs distribution. The solvable Markov random field model is equivalent to the Wiener filter or the simultaneous auto-regressive model. The estimates of hyperparameters are obtained by maximizing the marginal likelihood. We study the statistical property of the solvable Markov random field model, exactly.