HIERARCHICAL MAXIMUM-ENTROPY PARTITIONING IN TEXTURE IMAGE-ANALYSIS

This paper presents an effective texture representation which captures the statistics (or distributions) of structural and/or spatial relati ons of grey levels within certain neighborhood in a texture image. The structural and/or spatial relations are captured by various feature e xtraction operators to generate feature images. Then, the joint distri butions of the features which we termed feature frequency matrices (FF M) provide the statistics and representation of the texture image. A p artitioning scheme to 'compress' the FFM such that only relevant infor mation is retained is proposed. The partitioning scheme is based on th e hierarchical maximum entropy discretization scheme which minimizes t he loss of information. The efficacy of the representation is demonstr ated using homogeneous texture images.