Comparative analysis of texture descriptors for classification
Masood Ahmed, Arslan Shaukat, Muhammad Usman Akram · 2016
Texture analysis has been an important research area due to wide range of applications in the field of image processing, machine vision and pattern recognition. In this paper, we present a comprehensive analysis of texture descriptors for texture classification. We focus on state of the art texture descriptors which have been widely used for classification in literature and shown promising results. These descriptors comprise of Local Binary Patterns (LBP) and its various extensions, Gabor wavelets and fractal analysis which are invariant to scale, orientation and illumination. Some of the benchmark texture databases have been used for the extraction of features for each descriptor and then the classification performance was achieved using K-nearest neighbor (KNN) classifier. All texture descriptors performance is then compared and analyzed in detail. It has been observed that Local Binary Pattern generally outperforms other descriptors on three out of fours datasets.