Recent developments of the Markov Random Field (MRF) model, also known as the Gibbs Random Field, have shown that the MRF can provide a powerful image model for many image processing applications. In this paper the MRF model is applied to texture classification and texture discrimination problems which are formulated as statistical decision problems. Different texture classes are discriminated by using maximum-likelihood (ML) decision rules. The class-conditional likelihood functionals are derived from the conditional probability distribution function of the assumed MRF model. Results of texture classification and discrimination experiments are described.
|Title of host publication||Unknown Host Publication Title|
|Place of Publication||Princeton, NJ, USA|
|Number of pages||8|
|State||Published - Dec 1 1986|
ASJC Scopus subject areas