基于嵌入式系统结构的印鉴鉴别系统  

A Seal Imprint identifiction System Based on Embedded Structure

作　　者：刘铁根[1] 何瑾[1] 邓集杰[1] 朱均超[1] 

机构地区：[1]天津大学精仪学院光电信息技术科学教育部重点实验室,天津300072

出　　处：《光学学报》2009年第11期3043-3049,共7页Acta Optica Sinica

基　　金：国家自然科学基金(60627002)资助课题

摘　　要：为了弥补市场上现有印鉴鉴别系统体积大、移动性能差、安全性较低、价格比较昂贵等缺陷,研究了基于高速数字信号处理器(DSP)和现场可编程门阵列(FPGA)的嵌入式印鉴鉴别系统。系统在印鉴识别算法上,基于平滑卷积的方法计算印鉴的中心位置和半径。采用径向投影法,对一维特征数据计算,得到待验印鉴(SS)与预留印鉴(MS)之间的偏转角度。把印鉴质量指标作为MS与SS对应边缘Hausdorff距离测度的控制参数,用神经网络方法综合分析、判别印鉴真伪。在硬件实现方面,片上可编程(SOPC)系统结合DSP作为检测系统核心。SOPC系统包括控制器、图像预处理器等。DSP作为系统的主处理器,用于进行图像的特征检测与识别。系统具有以太网、RS232、USB等通用接口。实验表明,该系统可以有效识别印鉴,并具有体积小、成本低、系统功能可灵活升级等特点。An embedded seal imprint identifiction system based on high speed digital signal processor （DSP） and field-programmable gate array （FPGA） is designed and fabricated. In recognition algorithms of the system, spatial coordinates of centers and radius of tested seals are calculated with a method based on flatness convolution. The orientation between sample seal （SS） and model seal （MS） is calculated with the radius projection method, which calculates a group of one-dimensiond feature data instead of two-dimensiond. Improved Hausdorff distance is used to measure the similarity between seals, as seal quality being the control parameters. Artificial neural network is adopted in analyzing and estimation. In respect of hardware realization, System-on-a-programmable chip （SOPC） combining DSP serves as CPU of this verification system. Image preprocessor and controller are included in the SOPC system. As the master processor, DSP is applied to accomplish image feature detection and verification. General purpose interface （GPI） such as Ethernet interface, USB and RS232 is designed in the system. Experiment shows that seals can be effectively identified and the function of the small-size and low-cost verification system can be upgraded flexibly.

关 键 词：嵌入式系统 印鉴鉴别 HAUSDORFF距离 数字信号处理器 片上可编程 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]