关键词: 人脸表情识别, LBP特征, AVR特征选取, SVM分类器

Abstract: As the excellent capability of description of local texture，Local Binary Patterns（LBP） has been applied in many areas such as pattern recognition and computer vision.But the traditional LBP can not get good results in facial expression recognition.In this paper an improved LBP feature extraction method combined with wavelet decomposition is developed.First the face image is extracted and normalized through Adaboost face detection algorithm and 2D model，and then wavelet decomposition is applied to enhance LBP features，and the most effective features are selected through Augmented Variance Ratio（AVR） methods to reduce the dimension，and finally the Support Vector Machine（SVM） classifier is used to classify face images with enhanced LBP features.The experiment operated on JEFFE face image database proofs the effectiveness of the proposed method.

Key words: facial expression recognition, Local Binary Patterns（LBP） features, Augmented Variance Ratio（AVR） feature selection, Support Vector Machine（SVM） classifier