Research Of Face Recognition Algorithm In Intelligent Access Control System

With the progress of the society and information technology, the change of people’s ideology, the requirement of security systems is gradually increased, which makes intelligent access control system under the spotlight. Intelligent access control system that using face recognition for validation way is with the advantages of non-contact collection, suitable for hidden places, lower cost etc. The performance of intelligent access control system is directly affected by the identification technology. Above all, in this paper, based on some small-scale occasions application such as mini-office and home intelligent access control system, focus on the algorithms of face recognition to establishing face recognition system, and then carries on the test and analysis to lay the foundation for the application in the embedded system.According to the application environment and application platform, first of all, this topic discusses the development and research status of the access control system and face recognition technology at home and abroad, and considers the development trend of the access control system so that determining face recognition as the access control system authentication. Secondly, a number of algorithms of face detection are discussed, among all those algorithms, the basic concepts and algorithms of the face detection based on skin color model and AdaBoost algorithm for face detection are deeply researched. And a face detection algorithm based on skin color model and AdaBoost algorithm is proposed, which with the advantages of stronger robustness, low false detection rate and the miss rate by the combination of two algorithms. And then, the image preprocessing methods are analyzed in detail., using histogram equalization and median filtering to preprocessing the images those were successfully detected the human face in order to reduce light and noise that caused by the environment and facilities and it will improve the accuracy of face recognition in late period. Then, the related concepts and main methods of face recognition are briefly introduced, and the advantages and shortcomings of all kinds of face recognition algorithms are analyzed. PCA algorithm is selected to construct "face" space because it applies to small sample and with higher recognition rate, and the algorithm in the application of face recognition is researched, the selection of feature vector and distance classifiers are deeply analyzes, the Euclidean distance is used as the basis for face recognition. Finally, each part of the algorithm is determined, the feasibility of the algorithms is verified, and then the development of face recognition system based on OpenCV can be completed. The software environment for the development is Windows 7+VS2013+OpenCV 2.4.10, the performance of face recognition system is tested by a number of experiments, which including face detection module and face recognition module.A large number of experimental results are analyzed, ultimately, for static images, the average detection rate of the face detection module is 91.6%, the average error detection rate is 9.2%, for dynamic image, the detection effectiveness is good. The recognition rate of the face recognition module can achieve 91.7%. In general, the performance of the whole face recognition system is good.