On the effects of the Gibbs-Markov model parameters on MAP region segmentation using dynamic programming

The authors study the effect of clique potentials in the Gibbs-Markov model equations when used in maximum a posteriori (MAP) image segmentation of textured images. The focus is on the simple Derin-Elliott model in which the low level process consists of a constant gray level with additive Gaussian noise with zero mean and variance, and the high level process is a second order Gibbs-Markov random fields. Experimental results are presented