出版社：The Editorial Committee of the Interdisciplinary Information Sciences
摘要：For image restoration as an energy minimization problem, an energy function is expressed by using a hamiltonian of a classical spin system only with finite range interactions. The problem is formulated in terms of a probabilistic model with a Gibbs distribution and hence may be regarded as a Markov random field model. We compare the results obtained by proposed methods with those obtained by the probabilistic computational algorithms for restoration of natural images within a framework of the mean-field approximation or the pair approximation. We clarify a role of Nishimori-Wong’s inequality in the Markov random field model as for the practical algorithms to the image restoration of natural images through some numerical experiments.
关键词：Bayes statistics;probabilitic algorithm;quantum computing;Markov random fields;image restoration
