Design Of Palm Vein Collection And Recognition System Based On Zedboard

Recently, with the development of information technology and biotechnology, identity recognition technology using the physiological characteristics of inherent body has been widely applied in security, attendance and entrance guard system. Compared with the traditional biological recognition technology, palm vein authentication technology has received wide attention and research for its higher technical advantage of living recognition, organization characteristics, and security levels. For human hand vein hides under the skin, complex structure is difficult to be replicated, hand vein image is difficult to steal. The technique has the characteristics of uniqueness and stability, which makes the palm vein a kind of biological characteristics of the high security. Compared with the finger vein and the back of hand vein, palm vein contains more blood vessels and characteristics, which is better fit for feature identification. At present, Fujitsu palm vein recognition device is the most mature products, which has been widely used in ATM, entrance guard and security industry. In view of the commercial secrets, the details of the product realization are rarely described in the literature. Domestic research of vein collection and authentication system is in the infancy, most of the products extend its application with Fujitsu vein acquisition device, algorithm test validation dependents on the high-performance X86 extremely, independent research and development of palm vein authentication system are very few. In this paper we test and discuss mainly aims at light source selection, camera type hardware configuration, identification authentication algorithm for transplant.Hardware of palm vein identification and authentication system is mainly composed of collecting device and hardware platform. Acquisition device is made up of near infrared lamp and camera. For selection of light source and camera directly affects the quality of vein image, we chose mixed infrared source of 850nm and 940nm, distributed in rings. Besides, by comparing the quality of vein image acquired of different CMOS cameras, we determine a acquisition scheme of USB infrared camera composed of OV7720 image sensor chip. The test results show that the reasonable parameters on the sensor chip configuration benefits collection of hand vein image a lot. As a matter of fact, the realization of the algorithm depends too much on X86 hardware platform, in order to achieve small portable products, ARM and DSP embedded hardware are used as the test system in other references, and yet perform slowly on algorithm. In this paper, we choose Zedboard as the hardware platform for testing. As parts of processing system, CortexA9 completes image acquisition, image preprocessing, and one to one certification process. Programmable Logic completes two-dimensional discrete Fourier transform and inverse transformation operations to shorten Algorithm implementation time.Software implementation consists of hand vein image acquisition, preprocessing, feature extraction, and certification. Image acquisition include camera parameters configuration, original data acquisition and format conversion. Preprocessing includes binarization, edge extraction, selection of key point, extraction of ROI, and image enhancement. Palm vein ROI mainly includes center and thenar, in this paper, we choose the root points between fore finger and middle finger and ring finger as the key points, and the center of palm as the ROI. There are three kinds of features are used to extract, they are structural feature, texture feature and subspace feature. This article selects extraction of texture features and tests on different embedded platforms. We finally proposes a identification and authentication system of palm vein based on Zedboard after counting and comparing the time of algorithm implementation. Experimental results show that different part of Zedboard accomplishes different part of algorithm can allocate resources and shorten the working time, which meet the demand of real-time performance very well.Because of advantage of the technology, domestic and foreign research institutions and schools have done a lot of research. The first chapter summarizes this technology from software and hardware, and explains the content and meaning of the research. Firstly, the article details research background, compares different biological recognition technology and introduces the principle and advantage of palm vein identification technology.The second chapter gives an overview of the palm vein identification system. The content mainly includes the vein imaging principle, design of hardware structure and software realization process. Imaging principle part introduces acquisition mechanism, that cmos image sensor can acquire palm vein image with the method of illuminating palm by near infrared. The part of hardware structure is a brief introduction to the whole vein identification and authentication system, image acquisition device is composed of USB camera, system is controlled by CortexA9 processor on Zedbaord, FPGA implements algorithm acceleration. Software introduces process from collection to identification, including image acquisition storage, image format conversion, image pretreatment, identification and authentication.The third chapter details the hardware,which is the basic part of the system. Content mainly includes the introduction of light source, camera module, light filter, parameters of Zedboard, structure of the processing system, AXI interface, structure of FPGA, and circuit design. The most important part of circuit design details the System circuit, HDMI communication circuit, and the FFT operation circuit. The part of system circuit introduces the connection of overall circuit and the function of each module. HDMI communication circuit is responsible for LCD screen display. FFT operation circuit is responsible for completing Fourier transform and inverse transform for feature extraction.Software program design is the method to realize the entire system. The forth chapter introduces the process of software, including V4L2 framework application programming, BMP format image storage, vein image binarization, edge extraction, key points selection, ROI selection, feature extraction and certification. V4L2 framework application programming involves Camera parameters configuration, original data collection, and image storage. After the storage of grayscale images, we determine a specific threshold for binarization processing, so that we can get key points and edge extraction image, and then obtain the required ROI area for completing feature extraction and one to one certification.IP simulation result is the theoretical basis for the accelerated implementation of algorithm by FPGA. The fifth chapter introduces Fast Fourier Transform IP. The content includes FFT IP overview, a simulation program design, and the simulation results. FFT IP overview includes pin function, hardware structure, operation frame matching, IP timing and working instructions. Simulation program design introduces realization of FFT IP function by Verilog HDL. The part of simulation shows and analyses the experimental results.Appropriate operating system is the assurance for cooperation of hardware and software. The sixth chapter introduces Linaro Ubuntu operating system migration steps on Zedboard, to ensure the normal work of the system. Detailed steps involves making BOOT.BIN, zlmage, DTS, partitioning SD Card, and transplanting Linaro Ubuntu operating system. This chapter makes a detailed description of the above steps, through transplanting Linaro operating system to Zedboard, we build a complete software environment for the acquisition and certification system of palm vein.Experimental results part is verification of the research, and the sixth chapter analyzes the result of the experiment. Content is mainly divided into two parts, image acquisition contrast and time statistics of algorithm. The part of image acquisition compares the quality of palm vein image collected by different cameras of different configurations. Time statistics of algorithm part counts and compares the implementation time of all the parts of algorithm, verifying the rationality and efficiency of palm vein collection and recognition system based on Zedboard.In the final, the article gives a conclusion and outlook. Conclusion part summarizes the content and meaning of whole work, and briefly explains the work focus in this paper. At the same time the outlook part makes further explanation for deficiencies and problems of the system. Hopes and prospects are put forward to the further development and optimizing of the system.