Higher Order Models using Entropy, Markov Random Fields and Sequential Simulation

Abstract

Approaches to modelling using higher order statistics from statistical mechanics and image analysis are considered. The mathematics behind a very general model is briefly reviewed including a short look at entropy. Simulated annealing is viewed as an approximation to this model. Sequential simulation is briefly introduced as a second class of methods. The unilateral model is considered as being a member of both classes. It is applied to simulations using learnt conditional distributions.