Facial Expression Recognition Based on Local Binary Pattern and Gradient Directional Pattern

In this paper, we propose a new algorithm for facial expression recognition, which is based on gradient direction pattern (GDP), local binary pattern (LBP) and Sparse Representation Classification (SRC). The methods of gradient directional pattern and local binary pattern are used to extract features separately and then concatenate them as the final expression features. The Sparse Representation Classification is used to classify the test samples in seven categories of expressions. The experiment results based on Japanese Female Facial Expression (JAFFE) database demonstrate that this algorithm performances better than traditional methods such as LDA+SVM, 2DPCA+SVM etc.