Texture Classification via Attractive-and-Repulsive Decoded Gradient Contours

This paper presents a novel LBP-like descriptor for texture representation. The proposed method, referred to as attractive-and-repulsive decoded gradient contours (ARDGC), consists in dividing local features into two distinct categories, attractive and repulsive decoded gradient contours (ADGC and RDGC) thanks to the flexibility of attractive-repulsive characteristics of pixels in a 3 × 3 grayscale image patch. Unlike some existing methods which are based on pairwise comparison of adjacent pixels, the essence of the proposed ARDGC model is to encode the differences between local intensity values within triplets of pixels, along a route traced along the periphery of the 3 × 3 square neighborhood. In order to increase the robustness of the proposed operator, a new triplet, formed in addition to the central pixel, by the average local and average global gray levels, is incorporated in the modeling of ADGC and RDGC. The final multi-scale attractive-and-repulsive decoded gradient contours ARDGC descriptor is obtained by linear concatenation of ADGC and RDGC. Extensive experimental results on nine representative texture databases show that the proposed ARDGC descriptor demonstrates superior performance to 30 recent state-of-the-art LBP variants.