Improved Local Binary Pattern Based on a Novel Mapping Method for Texture Classification

Local Binary Pattern (LBP) has received widespread attention since proposed, especially in the fields of texture classification and face recognition. Most of the currently proposed LBP variants are still improved based on the traditional rotation-invariant uniform mapping method (LBP riu2 ). Nevertheless, LBP riu2 loses too much texture information indeed. Inspired by the idea of utilizing complementary information and multi-operator fusion, a new mapping method called Local Binary Pattern Based on Triple Complementary Information (LBP OUS ) is proposed in this paper. Compared with traditional and latest LBP mapping methods, LBP OUS achieves the highest classification accuracy on the Outex and CUReT datasets.