Centroid-based texture classification using the generalized Gamma distribution

This paper introduces a centroid-based (CB) supervised classification algorithm of textured images. In the context of scale/orientation decomposition, we demonstrate the possibility to develop centroid approach based on a stochastic modeling. The aim of this paper is twofold. Firstly, we introduce the generalized Gamma distribution (GΓD) for the modeling of wavelet coefficients. A comparative goodness-of-fit study with various univariate models reveals the potential of the proposed model. Secondly, we propose an algorithm to estimate the centroid from the collection of GΓD parameters. To speed-up the convergence of the steepest descent, we propose to include the Fisher information matrix in the optimization step. Experiments from various conventional texture databases are conducted and demonstrate the interest of the proposed classification algorithm.