Facial Expression Recognition Based on Binarized Statistical Image Features

This paper proposes a new algorithm for facial expression recognition based on a local feature descriptor which is used to extract binarized statistical image features (BSIF). Firstly, expression features are extracted by using BSIF descriptor. Then, the Sparse Representation-based Classification (SRC) method is used to classify the test samples in seven categories of expressions. We evaluate the performance of this method by classifying expressions in Japanese Female Facial Expression (JAFFFE) database. The experimental results show that our method improves accuracy in expression recognition tasks than traditional algorithms such as LDA+SVM, 2DPCA+SVM etc. The results testify the effectiveness of the proposed algorithm.