Image Texture Segmentation Using Local Binary Pattern and Color Information: A Comparison

Abstract

Image texture and color are important features for image segmentation. Several algorithms have been proposed using color features, texture features or combination of color and texture features for image segmentation in the literature. One of the important issues is how well these algorithms work on differentiating among different textures and colors. In this study, we analyze and compare some of the simple but powerful texture classifiers to explore their strengths and weaknesses in the classification of different type of textures. Texture Spectrum (TS) and Uniform local binary pattern (ULBP) are compared. In order to see the influence of color features in classification process, the combination of ULBP and color is also compared with ULBP and TS in our experiments. Co-occurrence probabilities (GLCPs) are used as a benchmark for the evaluation.