Research Of Access Control Management System Based On Face Image Recognition Technology

With the popularization of the computer application and the development of the network technology, people pay more and more attention on information security. People have developed a variety of methods in order to guarantee the security of information and improve the accuracy of identification. And along with the function ascension of smart phones, etc., the recognition technology that based on it becomes a hot research topic again, especially in some place that the flow of people is larger and the security requirement is higher. Compared to some of the common identity, face feature recognition has the advantages of out of contact, stability and uniqueness. so this thesis that combine face recognition with the embedded and mobile terminals, designs a entrance control system that is for high level of security.Face recognition system is divided into several parts, each part can influence the results of the whole system. In this thesis, the face detection,feature extraction and feature recognition step have been carried out with algorithm introduction, selection, integration, improvement, etc., which is applied to the smart phone platform and Windows platforms for face recognition system.In this thesis, some traditional face detection algorithm was studied, the algorithm that is suitable for this system is got through the selection and improvement of the existing algorithm. First preprocess the algorithm image through the skin color segmentation algorithm, divide the image to get a face that there may be, and then through the face length and area feature to selects the possible regional, finally through the Ada Boost algorithm to search the processing region, through the experimental verification, this thesis used the algorithm that combines the advantages of two algorithms, the algorithm can complement each other, improve the detection rate and improve the detection speed.In this thesis, after to compare the different feature points extraction algorithms, select to improve the sift algorithm for feature extraction and matching. The thesis describes the process of sift algorithm in detail. In the light of the main steps to improve. To realize the algorithm is a very complicated process, operation process of space and time complexity is high, a large amount of calculation is needed. The actual application to image processing is very difficult. The fourth part of this article through to improve sift scale space extreme point detection and character description vector, greatly reduces the calculation amount.A major innovation of this thesis is the access control management system, combined with Android platforms and embedded platforms to construct the face recognition system. Through mix together the detection and recognition algorithm in this thesis into the system, then realize the face detection and recognition system on the Android platform and embedded platform, and through the experiment to verify the function and performance index of the system.