A comparative study of Local directional pattern for texture classification

Abstract

Texture is an important feature for image analysis in which each pixel is classified based on its neighborhood. It is used for surface characterization in many applications, such as medical imaging, remote sensing and quality control. The purpose of this paper is to investigate the performance for the newly purposed Local directional pattern (LDP) and compared to the popular Gray level co-occurrence matrix (GLCM). In this paper, texture classification power of two feature methods, Gray Level Co-occurrence Matrix (GLCM) and Local Directional Pattern(LDP) are compared. Experiments are conducted on 25 Texture types selected from Brodatz album. Classification are carried out using 4 different classifiers (Naive-bayes(NB), Multilayer Perceptron(MLP), Support Vector Machine (SVM), k-nearest Neighbor Algorithm(k-NN)) in different conditions. In this study it is established that the LDP has the best accuracy at 97% using Multilayer Perceptron and 96% using SVM, compared to GLCM. In the literature Local Directional Pattern (LDP) has mainly been used to extract features in biometrics applications. In this paper LDP is used to characterize general purpose textures. It is shown that outperforms the very popular Gray Level Co-occurrence Matrix and Haralick features.