Smile Recognition Based on Face Texture and Mouth Shape Features

As an efficient feature descriptions in facial expression recognition, LBP was used to filter facial expression images in feature extraction because of its texture description and HOG was used because of its shape description. However, single feature description only reflects the unilateral feature information, ignores the relationship of various features and can't make use of diverse and complementary feature information. In this paper, a new smile recognition method is proposed based on feature fusion which combines the LBP and HOG features. Firstly, HOG features are extracted in the mouth region. And then facial expression features are presented with LBP histograms. Finally, the two feature set are fused to implement the smile recognition. The experimental results show that the proposed method can effectively utilize the global texture information and local shape information and get better recognition effect.