Facial Expression Recognition Based on Local Phase Quantization and Sparse Representation

In this paper, we propose a new algorithm for facial expression recognition (FER) based on Local Phase Quantization (LPQ) and sparse representation. Firstly, Features are extracted using LPQ descriptor. Then, Sparse Representation-based Classification (SRC) method is used to represent the test expression image by the linear combination of the training expression images. Facial expressions are distinguished by the residue analysis of sparse representation. The proposed algorithm is experimented on Japanese Female Facial Expression (JAFFE) database. The results show that the proposed algorithm is much better than those traditional methods, such as Local Binary Pattern (LBP) + Support Vector Machine (SVM), two-dimensional principal component analysis (2DPCA) + SVM, Linear Discriminant Analysis (LDA) +SVM etc. The performance is also improved obviously compared with the SRC algorithm. In addition, when images are under occlusion, recognition rate of the proposed algorithm also gets the highest recognition rate for FER.