Research And Realization Of Face Recognition System Construction Based On Anaconda-CL Board

With the rapid development of biometric technology, identification technology based on facial features has become a biometric field of research. Face recognition technology has many advantages, such as hidden high, fast, simple and accurate etc,and compared to traditional means of identification is more intuitive and easy to accept. With continuous research and development in many fields and disciplines, face recognition system has been widely used,such as meteorology, aerospace, criminal investigation, airport and station inspection,etc.Face recognition system principle is more complicated, cumbersome implementation process, in order to achieve the final effect of automated face recognition,face recognition requires many contentes and comprehensive evaluations.However, in practical applications, the majority of face recognition system using pure software or software-based, hardware-assisted way to achieve,its datas processing speed and recognition results is effective.If faced with a large number of image data and a more complex algorithm, it is difficult to ensure that the face recognition system to identify both good effect but also has a strong real-time.Therefore, how to ensure that the system has strong real-time recognition performance under the premise of face recognition system has become a key issue in the field of face recognition research.In this thesis, developed by the Canadian Dalsa dedicated image processing card--Anaconda card as system development and operating platform,Adaboost algorithm, Local Binary Pattern (LBP) algorithm has been converted into FPGA hardware language, use the hardware language for face detection and feature extraction,FPGA hardware with a pure face image in real time to detect and extract facial features, design complete solutions, supporting the development of functional modules for face recognition systems.Contents of this paper are as follows:(1) Detailed description of the recognition of the research background, describes the face recognition technology research and system construction, characteristics and development process, problems, etc.A brief description of the development trend of face recognition systems and applications;(2) Describes the Adaboost algorithm, Local Binary Pattern (LBP) algorithm,and based on the principle eigenvalues calculated distance and classification determination of human face, features, algorithm and other processes.(3) Anaconda card be used for achieving paper designed,a brief introduction to the structure of Anaconda cards, performance parameters, image capture and transfer case,PC section describes the basics of the characteristics Sapera LT library,base class library,basics of data. Describes the interface for data transmission in the hardware part.(4) Anaconda card-based face recognition system has been designed,and completed a face recognition experiment.Firstly the data source selection and data format have been introduced,then under the control of the host computer’s data transmission via on-board the bus to the image processing unit (FPGA), face detection and face igenvalues computing achieveed by hardware,then the data transfer to the PC,follow the instructions on the lower computer communication protocol separation valid datas after the host computere receives datas,categories face by nearest neighbor classifier after reading the database eigenvalues and basic personal information data,at last,face recognition system has been achived.This paper introduces the features of the face value for the extraction of facial feature extraction procedure used for system control and coordination of automated face recognition main program,its functional design and program flow are described.Through package design and implementation of real-time face recognition system,verify the feasibility of the program through a technical perspective.As at present, the entire development and design face recognition system has been basically completed.This system achieve the desired results,ensures a better recognition results while having some real time,provides a good reference for further research in real-time face recognition system.