<title>Image texture classification using quadrature mirror filter bank in combination with co-occurrence matrices</title>

Abstract

Abstract This paper presents a method for the classification of textures using Quadrature Mirror Filter (QMF) bank subband decomposition in combination with statistical descriptors. In our combined method the QMF bank splits the input image into four subbands, and statisti-cal descriptors based on co-occurrence matrices are computed from the subsanzpled low-low band. The experiments demonstrate that the combined method have better classification performance than that of statistical descriptors computed from the co-occurrence matrices of the whole texture image. In addition, the experiments demonstrate that the combined method based on computationally efficient IIR QMF banks yields approximately the sameclassification results as the combined method based on classical FIR QMF banks. 1 Introduction Texture analysis plays an important role in such areas as medical diagnosis, remote sensing and industrial inspection. One important usage of texture analysis is to distinguish among various classes of textures. That is, given a sample texture, identify to which of a finite number of textureclasses the sample belongs.For the texture discrimination problem several texture descriptors have been proposed: Fourierpower spectrum [1], autocorrelation function [2], autoregressive moving average models [3], Markovrandom field models [4], digital filter banks [5] and spatial gray-level co-occurrence matrices [6].Among the different texture descriptors spatial gray-level co-occurrence matrices are cited in theliterature most frequently. A co-occurrence matrix is a second-order statistical measure of imagevariation. It contains information on the spatial relationship between pairs of gray levels of pixels.Co-occurrence matrices are seldomly used directly. Instead, features based on them are com-puted. Haralick eL al. [6] have suggested 14 textural features, which include both statistical andinformation theoretic measures.