Research On Face Recognition Technique Based On Fisher Discriminant

It was reported that there are about 80% information is obtained from eyes. Due to the importance of human being in the multimedia information, the recognition based on man's biometrics information is one of the important topics in computer vision and pattern recognition. Face recognition is an active subject in the area of biometrics recognition. This interest is mainly motivated by the broad range of potential applications for systems able to recognize the face they contain. Examples include surveillance, personal identification, access control, conference, and human computer interface. Although human can recognition face and its expression with any efforts, face recognition is still a difficult task for computer, which is closed related to computer vision, pattern recognition, physiology, psychology and etc. The main challenges of face recognition lie in: 1) Face is a complex and high dimensional visual pattern. It needs dimension reduction and feature extraction in the processing of facial images. 2) Human image is 2D projection of a 3D object, in which scale, translation, rotation, pose, illumination, expression and etc are involved. Scale, translation and rotation are the common problems in image analysis; pose, illumination and expression are the particular problems in facial image processing. Research shows that there is greater variability in a given face across there three types of changes, than there is among different faces when these factors are held constant. This is a fundamental problem in terms of dimension reduction, feature extraction and face representation. 3) Since the facial image is high dimensional data, it may can not meet the real-time processing need in some applications, such as face detection/recognition in public places. 4) The number of classes is equal to the population in the database face recognition system, which maybe a very large number. Generally speaking, the recognition rate is dropped with the increase of classes. To the problems of face recognition, respectively, this dissertation reviews the existing theory and algorithm. Furthermore, it proves the old approaches. The main content of this dissertation is summarized below:...