Face Recognition Detection System Design And Research Based On Fpga

With the gradually increase of the science and technology requirement, andgradually increase of the security awareness, face recognition technology has becomeincreasingly wide range of community concern. The face recognition technology has theadvantage of simple operation, affordable, etc. Moreover, in recent years, the facerecognition technology has very good prospects for development in many areas, such asimage retrieval, immigration control, and intelligent human-computer interaction. And ithas showed a trend of rapid development.Face recognition technology is a mothed which analysis the human face use ofcomputer, and then to distinguish the identity which extracted from the image usefulidentifying information. Face recognition technology has broad application prospects.There are many algorithms of face detection. This article uses a combination of Adaboostalgorithm and particle filter, the two algorithms compensate each other. Although thespeed of detection is reduced, the inspection accuracy is raised greatly.The paper is mainly based on FPGA basis, application development platform isVirtex-II PRO (V2-Pro) development version. First of all, in recent years particularly fastpace of the development in the FPGA, it follows the development trend of the technology,the operation is relatively simple.Users can design their own module according to theirneeds, at the same time it left a lot of space for the upgrade and optimization of the users.FPGA technology isapplicated widely brcause its flexibility to reduce the cost of the costof the system design. Secondly, in the moment the product which produced by the XilinxVirtex-II family of products is used widely, and it has provided a guarantee for theDesign and Implementation of the face recognition system.The specific content of the paper is，(1)Algorithm analysis: Face detection is an important part of face recognitionsystem, face detection algorithm is to ensure that the protection of the testing carried outsmoothly. The system must ensure the accuracy of the detection rate and time-consumingalgorithm when analysis the algorithm. (2)Cooperation with software and hardware: the main consideration is now in themarket could provide resources situation, and the cost of the system. Therefore, the paperselected Virtex-II Pro development board which producted by Xilinx’s. It offers a PowerPC processor, USB2.0 interface, VGA output connection, bus and data channels, etc. Inaddition, the system still needs to speed up the implementation module.(3)The optimization of the cascade classifier training: the paper put this trainingmethod which can quickly improve the classification ability and reduce the characteristicnumber.(4)The integrity of the system to verify: Input and output of the image receiving,through experiments that the detection of the face recognition system speeds up to 25fps.And the system achieves a lot of detection rate and the relatively low by mistakeexamining rate.