Facial Expression Recognition Algorithm Research And System Implementation

Facial expression recognition is a kind of technology which computer determines facial expression and analyzes mental activity of a person according to prior knowledge in human emotional and affective computing. Facial expression recognition is the same as facial recognition technology, face images are processed. So there is a very close link between them. Facial expression recognition consists of reading image, image preprocessing, extracting feature and classification, especially extracting feature is the most important and hard. Facial expression recognition data is divided into two forms, which are static image and dynamic image. Because of different data form, there are also different algorithms to extract feature. Due to the libraries of dynamic image are few and library is rarely public, static image is used to process. The main contents implemented are as follows:1. Select facial expression database and detect face:make use of the biggest unicom region segmentation method to achieve face detection, and the detected face image is normalized by histogram equalization and scale normalization.2. Extract expression feature. Facial expression feature is respectively extracted by using local binarization algorithm, principal component analysis algorithm and Gabor wavelet transform algorithm.3. Classification. Support vector machine (SVM) algorithm is used to classify facial expression.4. Improved identification algorithm. Make use of the SVM to vote for expression feature by the way of one to one, which is extracted by using local binarization algorithm and principal component analysis algorithm. The two voting results are fused, and output the final result of facial expression recognition.5. Facial expression recognition system is designed and implemented. Make use of MATLAB to design the human-machine interface, to classify the facial expression.