Avatar Face Recognition Using Wavelet Transform and Hierarchical Multi-scale LBP

Recognizing avatars in virtual worlds is a very important issue for law enforcement agencies, terrorism and security experts. In this paper, a novel face recognition technique based on wavelet transform and Hierarchical Multi-scale Local Binary Pattern (HMLBP) is presented and shown to increase the accuracy of recognition of avatar faces. The proposed technique consists of three stages: preprocessing, feature extraction and recognition. In the preprocessing and feature extraction stages, the wavelet decomposition is used to enhance the common features of the same class of images and the HMLBP is used to extract representative features from each avatar face image without a need for any training. In the recognition stage, the Chi-Square distance is used to achieve a robust decision and to indicate the correct class to which the input image belongs. Experiments conducted on two manually cropped avatar image datasets (Second Life and Entropia Universe) show that the proposed technique performs better than traditional (single scale) LBP, Wavelet Local Binary Pattern (WLBP) and HMLBP in terms of accuracy (78.57% and 67.50% recognition rates for Second Life and Entropia Universe datasets respectively).