Facial Expression Recognition Research Based On Semi-supervised Learning Of People

With the rapid development and popularization of network, distance learning is becoming more and more extensive teaching approach. Compared to traditional face-to-face teaching, distance learning provides an effective way for full sharing of teaching resources, students in different regions can get the same teaching resources. However, this off-site teaching form realizes resource sharing, it also brings new problems, such as lack of emotional communication between instructors and students, leading to the effectiveness of teaching can not achieve the desired standards.Facial expression recognition technology has been widely used in the harmonious human-computer interaction, the technology through the computer to analyze the person’s facial expression and its changes automatically, the person being analyzed to determine the emotional changes and their ideological activities to achieve the human-computer interaction of natural intelligence. In particular, it has potential in the field of psychology of the application.This paper aims to in-depth study of facial expression recognition, and used in distance education to solve the drawbacks of distance education to make up for the lack of emotional problems in the process of distance education, so as to further improve the quality of distance education.Firstly elaborated the researching background, research status were analyzed and compared. By comparative and analysis, wavelet transform can effectively extract the characteristics of static facial expression image; semi-supervised learning can reduce the amount of marking, so this paper use the expression classifier based on semi-supervised learning for facial expression identification. Specific measures include the following aspects:first proposed the facial expression recognition technology will be applied to the network application background in education; in the realization of facial expression recognition in the process, we use binary images, geometric and gray level normalization obtained by a combination of facial expression areas; and multi-orientation Gabor wavelet transform two-dimensional face image feature extraction, maximum shielding and personal characteristics of the different lighting conditions; Finally, based on semi-supervised learning classifier used in facial expression recognition, reduce costs, and to obtain a better recognition rate.