Monte Carlo Markov chain techniques for unsupervised MRF-based image denoising

摘要: This paper deals with discontinuity-adaptive smoothing for recovering degraded images, when Markov random field models explicit lines are used, but no a priori information about the free parameters of related Gibbs distributions is available. The adopted approach based on maximization posterior distribution respect to line and parameters, while intensity assumed be clamped maximizer itself, conditioned parameters. enables application mixed-annealing algorithm maximum posteriori (MAP) estimation image field, chain Monte Carlo techniques, over binary variables only, simultaneous likelihood A practical procedure then derived which nearly as fast MAP reconstruction by known We derive method general case linear degradation process plus superposition additive noise, experimentally validate it sub-case denoising.