Facial expression recognition based on characteristics of block LGBP and sparse representation

摘要

Aiming at the difficulty of distinguishing texture feature in facial expressions recognition, this paper put forward a LGBP facial expression recognition algorithm based on the character blocking and sparse representation. The main contents are training the block images of different categories expression images and extracting the LGBP features of each sub-block. We construct a over-complete dictionary from which we get the discrepancy vector of each sub-block using sparse representation. Through finding the minimum residual vectors to achieve the recognition of different facial expressions. To some extent, the experimental results based on the JAFFF and Cohn-kanade facial expression database show that this algorithm can effectively overcome the influence of texture feature differences and have higher recognition rate.