A Facial Expression Recognition Method Based on Feature Blocks

Abstract

In this paper, a novel method for human facial expression recognition (FER) is proposed. It adopts the idea that the appearance of a region of interest can be well characterized by the distribution of its local features. Considering the importance of the eyes and mouth for FER and the outstanding performance of local binary pattern (LBP) to extract local textures, a representation model for facial expressions based on feature blocks and LBP descriptors is proposed. The strategies of FER including face normalization, feature-block acquisition, and LBP feature extraction are explained in detail. A principal component analysis (PCA) method is implemented to learn the structure of the expression in the LBP feature space. A recognition experiment is conducted on the JAFFE facial expression and TFEID databases using the nearest neighbor classifier. Experimental results confirm that the method is simple and demonstrates competitive performance.