Face Recognition Technology Research Based On Facial Character

Human face recognition technology has very high academic and practical values. It is attractive in pattern recognition and image processing. Face recognition technology will become the most potential method of identity recognition because of the particularity of human faces. Recently, with the development of society and the improvement of science and technology, the actual demand of convenient and reliable automatic recognition of human face is urgently needed. And its extensive application is became the most concern issue of numerous researcher of computer scientific. Although there are many algorithms of face recognition in the past years, the rate and speed of discrimination, Furthermore, the auther also establishes a face recognization simulation system which including image pre-processing, face detection, feature extraction and recognition, the effect of these algorithms are unreached anticipated purpose.In this paper, Collecting and learning many essays, researching papers related human face detection and recognition of the domestic and international in recent years, discussed some method about face detection and face recognition. And contraposing the related method of an automatic face detection and recognition, we do some researches deeply. We also probe into facial feature extraction. We propose an improved method of face recognition, combining the feature of front and profile to identify status. Experiments indicated that the methods of face detection and face recognition in this paper considered reasonable, which have certain academic value and practical value.In this thesis, we fisrt introduced the content and the difficult of face recognition. In order to detect human face, we introduce some typical color model fistly in this paper. After the lighting compensation, we use a method which combined skin color detection and searched for the apple of eye to detect the area of face. The distribution and characteristic of skin color is firstly analyzed in color model. Adopt an effective skin color model in face image segmentation. Then, locate the apple of eye to find the face region. The experiment results show that this method improve the detection rates.In the part of feature extracting and recognition, we use one of the most popular methods to classify human face. And a new algorithm based on frontal and profile feature was proposed. We probe into constructing the model of profile nose when we extract the profile feature. Use the characteristic of nose to improve the speed of identification contrast and storage. Experiment indicated that the model has some reference value. We adopt a method of order scoring and eigenvalue in the feature matching, improve the rate of facial recognition.At last, a face detection and recognition simulation system is established. The experiment results show that it robust to the face image database constructed by ourselves.