Design Of Intelligent Entrance Guard System Based On Face Recognition

In the high-speed development of the society, security equipment get more and more people's attention, Appeared in a variety of access control equipment. At the same time, the biological recognition technology has made great development, such as fingerprint recognition, iris recognition and face recognition has been widely used in people's real life. The face recognition is widely used in the intelligent entrance guard system because of it is easy to use, difficult to counterfeit, high recognition rate and so on. Most face recognition entrance guard system is built on the basis of PC currently, with high cost, large size, such as defects. In recent years, embedded technology rapid development, To realize face recognition and integrated into the security system on the embedded devices has become possible.In order to combine facial recognition technology with embedded devices, and used in intelligent access control systems of certain fixed occasions in the end, we put forward that use the ARM9 as core processor, the operating system is Linux,use PCA as face recognition algorithm, intelligent mobile phone for remote control solution, design an intelligent access control system on embedded devices. In the thesis, include hardware design, software design, algorithm design and other aspects to elaborate solutions for intelligent access control system based on face recognition.hardware design based on embedded development board which is product in tianqian, using the chip of S3C2440 as the core processor, using CS8900A as network communication module, using high-definition digital camera with a USB interface as video image acquisition devices, peripheral control circuitry design with a tensile 180KG magnetic locks for access control.Software design include the building of development environment, such as the kernel, file system migration, Qt compiling and transplant, opencv library to compile and transplant, access control lock device driver design, the camera image acquisition program design, face detection and recognition program designed library remote programming network requests, create a applications of Qt to provide users with a good man-machine interface.The altorighms includes image pretreatment, face detection and recognition, the pretreatment contains motion detection, image dimension reduction, image geometry normalization, image equalization and median filtering algorithm. Face detection section compares the features of geometric and template-based feature detection algorithm, finaly use the Adaboost algorithm cascade classifier for face detection, detection algorithm inorder to accelerate extract haar-like features using the integral value of the diagram, and use Adaboost algorithm to train each single weak classifier training into a single strong classifier, the last formed multiple strong classifier into face recognition library. Face Detection for face-targeting methods based on biometric identification, Face image cropping and normalize operation, provide a source for the subsequent image recognition training and recognition. By comparing the identifying portion based on knowledge, template matching and feature recognition algorithms neural networks, the design choose of PCA algorithm for face images of training and recognition, and recognition results shared to other applications.The results showed that this design achieved the face recognition system based on ARM, It can control electromagnetic locks for access control, face recognition subsystem reached more than 90% of the face detection accuracy rate and more than 85% of the stranger alarm Accuracy. Embedded face recognition system with respect to the PC with low cost, low power consumption, better scalability, and other characteristics, for a small number of people out of the case basically reached the real-time, reliable and secure access control system requirements.