Facial Expression Recognition Research And Analysis

Facial expression contains plentiful information of human behavior. In human society, facial expression is the most important communication method besides human voice. As a carrier of information, expressions can convey lots of information that voice can not compare. The facial Expression recognition extracts the features of human facial expressions information, then to classify and synthesis them according to the human cognition and consideration and making use of the known information of human's emotion, which makes computer can do some intelligent consideration in some extent and analysis the emotion of human. The system of facial expression recognition usually contains three basic parts: preprocess of facial expression images, the extraction of facial expression features, and the recognition of facial expression.In this paper, we review the history and the development of facial expression; study the methods of facial expression recognition by the way of pattern recognition and statistic learning. Firstly we adopt the feature extraction algorithm based on the combined Gabor wavelet transformation and fiducially point. Secondly we propose a facial expression method based on GMM, which uses the Gabor wavelet transformation as the input feature and get some meaningful result. Finally the paper also compares and analysis the classification ability of SVM and GMM. The experience shows the effectiveness of GMM in human facial expression, but we also could conclude that the SVM could be better in the result of classification.