The Research About Human Face Detection And Recognition

Human face recognition is of great importance in the field of theory and practice. It can be applied to security system, human I.D. management, and so on.In this paper, human face recognition is discussed from several parts which include human face detection, pretreatment, feature extraction, the design of the classified method.During the course of human face detection, the comparability of color information is applied. As the pretreatment, at first, the eyes are positioned through the template matching and projection. For the unification of size of face, space between two eyes serves as standard. In this paper, the nearest interpolation and double linearity interpolation are debated comparably during the course of the unification.In the feature extraction step, the procedure of eigenfaces construction through K-L transform is discussed. The SVD(single value decomposition) is enforced during the course of eigenfaces construction. As the second method of feature extraction, fisher LDA(linear discriminate analysis) is analyzed systemically, then the projection matrix is gained.In this paper, the K-L transform and fisher LDA are realized with the ORL databases. For the K-L transform, the classified information is concentrated on the principal component. Through the fisher LDA, the effective projection orientation and better recognition rate are gained.