A Research On Face Recognition Technique Based On Feature Points

With the advent of the information age, people more and more faced with the problem of identity authentication in their daily lives. Biometric identification technology as a new identification method, can overcome the shortcomings of traditional methods, and is more secure, reliable, accurate and convenient. Biometric identification technology refers to using the human inherent physical characteristics or behavioral characteristics to carry personal identification by computer. Face recognition was the most acceptable method of authentication, because it had features of non-invasive, direct, friendly, convenient, and did not require special equipment. Therefore, the face recognition became the hot spot in biometric identification technology.At present, there are many ways of face recognition at home and abroad. These ways can get relatively high recognition rate in the controlled observation conditions. As the effects of light, gesture, expression, and material, people studied the face recognition in particular contexts, also raised a number of fast and efficient algorithms for different environments. Dye spot and dye birthmark are referred to as "feature point". While the relatively small number of these people, but in real life, after all there are some people. Particularly, such groups often encounter in the period of thesis research. From the people's prior knowledge, these characteristics for identifying a person's identity is very important, because these feature points is a necessary condition for face recognition. If a person without feature points can be quickly ruled out, so feature points have an important role in access control, security, defense, public security and criminal investigation. So we have a need for the type of people to study a rapid and efficient approach for face recognition. The purpose of this research is to find a way to effectivly use the facial feature points for fast and accurate face recognition.In the image pre-processing stage, firstly image size and gray-scale must be equal. Secondly, highlitht the important facial image feature points by extending gray. Finally, face characteristic is partitioned in the method of threshold or area-growth.In the stage of feature extraction, according to the transcendent knowledge of face characteristic, three methods: shape feature, local PCA, wave transform are studied. We compare with these three methods. It is known that they have common ground in decreasing the dimension and keeping down the main feature of image. The shape feature method is robust in face gesture. But it ignores the other main information in the face image. So ti is only fit to the rough classification. Local PCA is uncertainty in division. Furthermore, it is complex in calculating. Wavelet transform can extract the local information from image. Furthermore, it is robust in lightness and simple in implement. Thinking over these three methods, this text uses wavelet transform in expressing face image. Base on wavelet transform, feature points is used identifiable pixels and extracted local and global Gabor feature.In the phase of classification, wavelet neural network approach is adopted. Wavelet neural network has inherited advantages of both wavelet and neural network. Through the training of adaptive wavelet basis to adjust the shape of the realization of wavelet transform, and has a good ability to function approximation and pattern classification capabilities. In the same time, taking full advantage of wavelet transform time-frequency localization characteristics of the subtle characteristics of the signal extraction improved frequency localization properties and multi-scale features of wavelet neural network.In this paper, features fusion method of the combination of global and local features is adopted for face recognition. Local Gabor feature vector and global Gabor feature vector combined and will be as a wavelet neural network input layer node vector. Experiments show that this method can effectively identify a person with facial feature points.