UNSUPERVISED TEXTURE SEGMENTATION USING MARKOV RANDOM FIELD MODELS

We consider the problem of unsupervised segmentation of textured images. The only explicit assumption made is that the intensity data can be modeled by a Gauss Markov random field (GMRF). The image is divided into number of nonoverlapping regions and the GMRF parameters are computed from each of these regions. A simple clustering method is used to merge regions. The parameters of the model estimated from the clustered segments are then used in two different schemes, one being an approximation to the maximum a posteriori estimate of the labels and the other minimizing the percentage misclassification error. Our approach is contrasted with a recently published algorithm [1] which detailed an interesting simultaneous parameter estimation and segmentation scheme. We compare the results of the adaptive segmentation algorithm in [1] with a simple nearest neighbor classification scheme to show that if enough information is available, simple techniques could be used as alternatives to computationally expensive schemes.