指纹图像比对算法研究及SOPC实现

Study on Fingerprint Image Matching Algorithm and SOPC Implementation

作者: 专业:电路与系统 导师:戚金清 年度:2010 学位:硕士  院校: 大连理工大学

Keywords

Fingerprint Identification, FMT, SOPC, FPGA

        指纹识别是生物识别领域中最成熟的鉴定技术,在金融,公安,防伪等领域得到了广泛的应用。本文在现有的研究成果的基础上改进了一种基于相位相关性的指纹识别算法,同时在ALTERA DE2开发板上进行了实现。本文首先研究基于相位相关性指纹识别算法存在的问题,在此基础上提出了一种改进的算法。在该算法中,运用傅里叶梅林变换算法克服图像间的相对平移、旋转和缩放,同时考虑到相关图形中的峰值分布特点和傅里叶变换的中心对称性,对相关图像进行镜像操作。判别条件不仅依赖峰值的对比度而且考虑了主次峰的位置关系。通过实验证明:与BLPOC算法相比,该算法使EER降低了3%。其次搭建了基于FPGA(SOPC)的指纹识别系统。该系统中指纹图像的缓存、图像增强和数据传输接口是由FPGA的硬件模块实现,而指纹的傅里叶梅林变换和比对过程则是在NIOS II软核中实现。并且该系统中的SD卡用来存储指纹模板库。SD卡挂接在Avalon总线上,通过SPI协议传输数据,并成功移植了ZLG文件系统。由于NIOS II软核中图像的二维傅里叶变换速度慢,所以本论文对傅里叶变换进行了硬件加速,使得指纹图像傅里叶变换速度提高了13倍。最后分析指纹识别系统中指纹模板存在的风险性,研究了伪随机数和指纹模板结合的双因子认证方法。通过Matlab的实验,得出了双因子算法中起主要作用的是伪随机向量组,增加伪随机向量组长度可以提高算法的可行性。
    Fingerprint recognition is the most mature technology in the field of biometrics identification and it is widely used in the field of finance, public security, security and other fields. Based on the current study results, the classification for fingerprint recognition algorithm is done in paper. It focuses on the key improvement of the phase correlation algorithm based on FMT. The algorithm is implemented on the ALTERA DE2 development board.Firstly, the problem of phase correlation-based fingerprint recognition algorithm is studied and an improved algorithm is proposed. The algorithm makes full use of the spectrum information of fingerprint image to do the fingerprint matching. In this algorithm, the use of Fourier-Mellin transform algorithm overcomes the relative shift, rotation and scaling between images. Meanwhile, taking into account the relevant peak in the distribution graph and the center of symmetry of Fourier transform, mirror operation is adopted for the relevant images. Our motivation is that not only dependent on the contrast of the peak but also considers the relationship between primary and secondary peak position. In addition, experimental results show that EER is reduced by 3% compared with BLPOC.Secondly, the Fingerprint Identification System based on SOPC is built. In the system, cache of fingerprint images, image enhancement and data transfer interface are implemented by FPGA while the Fourier-Mellin transform and verification are implemented in the NIOSⅡsoftware core. In addition, SD card is adopted to store the fingerprint templates and it is attached to the Avalon bus to transfer data via SPI protocol, mean while, ZLG file system is transplanted successfully.Thirdly, in the fingerprint identification SOPC system, the software part of Fourier transform consumes lots of resources and affects the overall performance. C2H technology is employed and the processing speed is improved 13 times faster.Finally, two-factor authentication algorithm is studied in this paper. Experimental results show that the pseudo-random vector plays a major role in the algorithm. The length of vector can be increased to ensure feasibility of the method.
        

指纹图像比对算法研究及SOPC实现

摘要4-5
Abstract5
1 绪论8-12
    1.1 引言8
    1.2 自动指纹识别系统研究现状和意义8-10
    1.3 指纹特征模板保护10
    1.4 本文组织结构10-12
2 改进的指纹识别算法12-27
    2.1 识别算法的评价指标12-13
        2.1.1 拒判率和误判率12-13
        2.1.2 指纹算法的鲁棒性13
        2.1.3 识别模式分类13
    2.2 相关性指纹识别算法理论13-16
        2.2.1 POC算法13-14
        2.2.2 FMT变换基本原理14-16
    2.3 相关性算法分类及存在问题16-17
    2.4 改进指纹识别算法17-23
        2.4.1 指纹图像的预处理17-19
        2.4.2 相位相关19-20
        2.4.3 镜像操作20-21
        2.4.4 峰值对比21-23
    2.5 实验结果23-26
    2.6 本章小结26-27
3 指纹识别SOPC系统硬件设计27-44
    3.1 系统框架设计27-28
    3.2 图像增强模块28-32
    3.3 图像缓存模块32-34
    3.4 LCM控制器模块34-38
        3.4.1 LCM配置模块35-36
        3.4.2 LCM数据接口模块36-38
    3.5 NiosⅡ系统硬件设计38-39
    3.6 自定义接口设计39-43
        3.6.1 SRAM控制器接口39-41
        3.6.2 数据传输接口41-43
    3.7 本章小结43-44
4 系统软件设计及实现44-49
    4.1 简介44-45
    4.2 软件部分流程图45
    4.3 SD卡程序设计45-48
    4.4 本章小结48-49
5 关键模块优化设计49-57
    5.1 傅里叶变换的硬件加速49-55
        5.1.1 快速傅里叶变换简介49-50
        5.1.2 硬件加速简介50
        5.1.3 硬件加速50-54
        5.1.4 功能验证54-55
    5.2 自定义指令55-56
        5.2.1 自定义指令简介55
        5.2.2 定制指令设计实例55-56
        5.2.3 功能验证56
    5.3 本章小结56-57
6 指纹模板保护算法研究57-62
    6.1 双因子认证简介57-59
        6.1.1 指纹模板保护算法的流程57-59
        6.1.2 汉明距离59
    6.2 实验结果59-61
    6.3 本章小结61-62
结论62-63
参考文献63-65
攻读硕士学位期间发表学术论文情况65-66
致谢66-68
        下载全文需10


本文地址:

上一篇:H.264编码标准中帧内预测及全零块预判算法研究
下一篇:恒温炉内铸件温度无线检测系统设计

分享到: 分享指纹图像比对算法研究及SOPC实现到腾讯微博           收藏
评论排行
公告