Texture Image Classification Using Gabor and LBP Feature

This paper presents a feature fusion based texture image classification method simultaneously using Gabor and Local Binary Patterns (LBP) feature. LBP and Gabor wavelets are two widely used two successful local image representation methods. This paper proposes two kinds of feature fusion methods, which perform in feature level and matching score level, respectively. We show that combining the two successful local image representations, i.e. Gabor wavelets and LBP, gives considerably better performance than either alone. Experiment results on MIT texture database demonstrate the effectiveness of our method.