Face Recognition Algorithm Based on Multi-layer Weighted LBP

Since the basic LBP operator rigidly field point value of 1 and 0, there exist some error that different face images with the same LBP code value, a face recognition algorithm based on hierarchical weighted local binary pattern (MW-LBP) is proposed. Firstly, the original face image is divided into small blocks, comparing the neighbor pixel values and the center pixel values at each block, using multi-layer weighted LBP feature algorithm to extract histogram of each block respectively, and then keep the partitioned histogram cascade at all levels to form the hierarchy histogram, the every level histogram series into a general histogram vector. Finally, the classification is performed using a nearest neighbor classifier with Chi square as a dissimilarity measure. The proposed algorithm experiments on ORL and YALE face database, experimental results show that MW-LBP algorithm can obtain a better recognition rate and has good recognition effect when comparing with original LBP algorithm.