RGB color distribution analysis using volumetric fractal dimension

Abstract

Over the years many approaches for texture analysis have been proposed. Most of these methods use, directly or indirectly, the spatial information to build the features. Although the spatial distribution of gray levels is a property a priori of the texture, some methods do not use this propriety to characterize it. The problem is that this class of methods has, generally, worst results than first one. Thus, in this work we propose a new method to classify color textures that does not use any type of spatial distribution information and still achieves high classification rates, comparable, if not better, than traditional texture analysis methods. The method is based on analysis of RGB color distribution using volumetric fractal dimension. © 2012 Springer-Verlag.