Improving Potts MRF model parameter estimation using higher-order neighborhood systems on stochastic image modeling

This paper presents a novel pseudo-likelihood equation for the estimation of the Potts MRF model parameter on third-order neighborhood systems, allowing the modeling of less restrictive contextual systems in a large number of MRF applications in a computationally feasible way. The evaluation is done by a hypothesis testing approach using our approximation for the maximum pseudo-likelihood (MPL) estimator asymptotic variance. The test statistics together with the p-values, provide a complete framework for quantitative analysis in MRF parameter estimation on stochastic image modeling.