SEGMENTATION OF GABOR-FILTERED TEXTURES USING DETERMINISTIC RELAXATION

A supervised texture segmentation scheme is proposed in this article. The texture features are extracted by filtering the given image using a filter bank consisting of a number of Gabor filters with different f requencies, resolutions, and orientations, The segmentation model cons ists of feature formation, partition, and competition processes, In th e feature formation process, the texture features from the Gabor filte r bank are modeled as a Gaussian distribution, The image partition is represented as a noncausal Markov random field (MRF) by means of the p artition process, The competition process constrains the overall syste m to have a single label for each pixel, Using these three random proc esses, the a posteriori probability of each pixel label is expressed a s a Gibbs distribution, The corresponding Gibbs energy function is imp lemented as a set of constraints on each pixel by using a neural netwo rk model based on Hopfield network, A deterministic relaxation strateg y is used to evolve the minimum energy state of the network, correspon ding to a maximum a posteriori (MAP) probability, This results in an o ptimal segmentation of the textured image. The performance of the sche me is demonstrated on a variety of images including images from remote sensing.