This paper deals with markov random field and it's application in image texture classification.We concerned with three problem:choice of an appropriate model from a class of such competing models;estimation of the unknown paraneters in markov random field models and using parameters to classify images.For the first problem

we use bayes decision rule;for the second problem

we use maximum likelihood estimation methods;for the third problem

we use fuzzy assemble analysis. At last

we did some tests.