Design And Implementation Of Security System Based On Face Recognition

With the development of technology and society, some of the traditional means of identification, such as IC card security, passwords, etc. has become increasingly unable to meet people’s needs for security. So the research and development of biometric also seems logical. In this paper, the design and implementation of a security system which is based on face recognition is used to control the work of the park entrance. The system can help workers control the information of people who had visited in a convenient and efficient way. Compared to other access control system, the system which is base on face recognition technology is more friendly and fast.The main work is as follows:First, we introduce some basic face recognition technology, for some current mainstream face detection and face recognition algorithm made a statement and research.Secondly, we give a short introduction of security system, and gives an overview of the system design.Then we have to carry out research in the field of face detection. The main research is the face detection which is based on color, including color space conversion, color modeling, etc. and for a simple face detection algorithm based on skin color to do the research. We are also have done some research and presentation about a number of techniques for image preprocessing. Based on the above study, we propose a fast, adaptive color-based face detection algorithm. Implementation is convert RGB color space to HSV color space, to extract color histogram and build color model, and then histogram projection, image morphology operation to eliminate interference. The results show that this method is reliable.Then face recognition. This paper uses face recognition method based on geometric features. In the paper, we made a series of feature region comprising the initial location, feature extraction, feature vector configuration and feature matching. The main idea is projected by pretreatment good gray face image, to determine the initial area. Then determine the precise area based on the geometric characteristics of the organ and the variance of the projection and the use of edge detection technique to extract feature points, according to the structural features characteristic point vectors, and finally using the weighted similarity matching methods.Finally, we tested the system, the test results show that our algorithms and systems are reliable and feasible.