摘要
纸币清分机是一种集光机电磁于一体的高端金融现金处理设备，其主要功能是对纸币进行面额、面向、新旧程度的识别，以满足挑选适合ATM取款机纸币、代替人工高效地选择可再流通的纸币等要求。随着模式识别理论在近三十年内获得高速的发展，通过光电图像来识别纸币的计算机技术于20世纪90年代开始陆续出现，并在纸币清分机上得以应用，使识别率进一步提高。早期采用数字图像识别技术的清分机由于受到芯片处理能力的限制，存在运算能力低下、机械结构复杂、体积庞大等缺点。21世纪初随着高性能数字信号处理器的出现，使清分机小型化、高速度的目标成为可能。为了打破国外清分机对我国金融市场的垄断，以最新的芯片技术研发性能优良且价格合理的清分机成为国内高端金融设备研发商的重要目标。
本文首先综述了当前国内外纸币清分机的研发现状，构建了一套由图像处理电路(包括主运算DSP芯片TMS320C6414和外扩存储器)、输入通路(包括A／D转换器、缓冲器)、逻辑控制器、调试端口(JTAG，Joint Test Action Group联合测试行动小组)、电源管理等各部分综合而成的智能纸币清分机光电图像处理平台，并对包括接触式图像传感器(CIS，Contact Image Sensor)等重要器件进行了选型分析。
系统的核心是TI公司的TMS320C6414 DSP芯片，通过CPLD (ComplexProgrammable Logic Devices复杂可编程逻辑器件)逻辑控制，利用该DSP强大的数据处理能力、高达1024 KB的内部Cache／RAM结构以及丰富的集成外设资源，能完成纸币图像处理算法的各项任务，满足清分机苛刻的实时性要求。系统硬件具有两个在该领域的创新点，一是首次将高性能DSP芯片TMS320C6414应用到清分机系统上，通过EMIF(External Memory Interface，外部存储器接口)进行大容量存储空间的扩充，保证了识别人民币需要的大数据容量要求。二是根据图像传感器的四通道输出结构，设计了相应的拼接式数据采集模式，保证了数据采集的完整性。这两个创新点结合起来，使清分机变得更轻巧，速度更快，性价比更高。此外，还研究了系统的程序加载过程，编写了适合本系统的FLASH存储器烧写程序及程序自动加载程序。
Currency sorter is an advanced financial processing device integrated with optics, mechanics, electronics and magnetics. Its main function is to recognize the denomination and face direction of the bills and distinguish the newer bills from the older ones. Henceforth, currency sorter may satisfy the general operation on ATM such as drawing out bills and the selection of reusable bills automatically. With the high-speed development of the theory of pattern recognition in the last thirty years, the computer technology of bill's photoelectric image recognition which used on currency sorter to improve its recognition ratio, has appeared in succession in the last decade of 20 century. The early currency sorter which used the digital image recognition technology is restricted by capability of CPU. They have some shortages: poor calculation skills, complex mechanism, bulky machinery and so on. With the invention of high-performance DSPs, the goal that minimize and quicken current sorter simultaneously has became more feasible. To break monopolization of foreign currency sorter, the advanced financial facility company instituted an important goal to research and develop currency sorter which used the streamlined CMOS chip technology and it has both good performance and low price.
Firstly, the current status of the bill sorter development is summarized, then a system has been established which consists of five parts, that is, image process portion(including TMS320C6414 and memories),input channel(included contact image sensor, analog-digital converters and buffers),logic controller, debugger(JTAG) and power manager.
The key component of this system is the DSP chip TMS320C6414 made in TI. The logic is controlled by CPLD. This DSP owns powerful capabilities of processing datum, 1024 KB internal Cache/RAM structure and abundant resource of integrated peripherals. It can execute varieties of tasks, which are required by bill image processing algorithm, and is fit for the rigorous demand of real-time. There are two innovations in the hardware system in currency sorter field. One of the innovations is that the TMS320C6414 is applied on currency sorter system for the first time. This project
引文
[1] F.Takeda, S.Omatu, S.Onami. Recognition System of US Dollars Using a Neural Network with Random Masks. Proceedings of 1993 International Joint Conference on Neural Neworks. 1993: 2033-2036
[2] F.Takeda, S.Omatu. High Speed Paper Currency Recognition by Neural Networks. IEEE trans on neural networks, 6(1), 1995:73-77
[3] F.Takeda, S.Omatu, S.Onami, et al. A Paper Currency Recognition Method by a Small Size Neural Network with Optimized Masks by GA. IEEE trans on neural networks, 8(4), 1994: 4243-4246
[4] F.Takeda, S.Omatu. A Neuro-Paper Currency Recognition Method Using Optimized Masks by Genetic Algorithm. IEEE trans on neural networks, 6(3), 1995: 4367-4371
[5] M.Tanaka, F.Takeda, K.Ohkouchi, et al. Recognition of Paper Currencies by Hybrid Neural Network. IEEE trans on neural networks, 8(1),1998: 1748-1753
[6] F.Takeda, T.Nishikage. Multiple kinds of Paper Currency Recognition using Neural Network and application for Euro Currency. IEEE trans on neural networks, 7(6), 2000: 143-147
[7] A.Ahmadi, S.Omatu, M.Yoshioka. Implementing a Reliable Neuro-Classifier for Paper Currency Using PCA Algorithm. SICE, 20,2002: 2466-2468
[8] A.Ahmadi, S.Omatu, T.Kosaka. A Reliable Method for Recognition of Paper Currency by Approach to Local PCA. IEEE trans on neural networks, 6(2), 2003: 1258-1262
[9] A.Ahmadi, S.Omatu, T.Kosaka. A Methodology to Evaluate and Improve Reliablility in Paper Currency Neuro-Classifiers.Proceedings 2003 IEEE International Symposium on Computational Intelligence in Robotics and Automation July 16-20, 2003, Kobe, Japan
[10] Angelo Frosini, Marco Gori. A Neural Network-Based Model for Paper Currency Recognition and Verification. IEEE trans on neural networks, 7(6), 1996: 1482-1490
[11] Jae-Kang Lee, Hwan Kim. New Recognition Algorithm for Various Kinds of Euro Banknotes. IEEE trans on neural networks, 7(6), 2003: 2266-2270
[12] Jae-Kang Lee, S.Goo Jeon, Hwan Kim. Distinctive Point Extraction and Recognition Algorithm for Various Kinds of Euro Banknotes. International Journal of Control,Automation,and Systems, Vol.2, No.2, June 2004
[13] 刘家锋，刘松波，唐降龙。一种实时纸币识别方法的研究。计算机研究与发展，Vol．40, No．7, July 2003
[14] 罗皓斐。纸币清分机图像处理系统的研究。浙江大学硕士学位论文，2005．2
[15] F. Takeda, T. Nishikage, Y. Matsumoto. Development of Autonomic Neural Board for Banknotes and Advancement to Palm Prints Recognition. IEEE trans on neural networks, 3(2), 1999: 3246～3250
[16] F. Takeda, T. Nishikage. Development of Autonomic Neural Board and Advancement to Palm Prints Recognition. IEEE trans on neural networks, 2(4), 2000:160～164
[17] Texas Instruments Co. TMS320C6414T, TMS320C6415T, TMS320C6416T Fixed-point Digital Signal Processors, July, 2004
[18] Syscan Co. SV255A4H DATA SHEET, 2003
[19] 冯涛，曹丹华，吴裕斌。基于DSP的CIS信号采集与处理技术。测控技术，Vol．22，No．8，2003
[20] 胡广书。数字图像处理——理论、算法与实现。清华大学出版社，1997
[21] AD9059 DataSheet. Analog Devices Incorporated, 2003
[22] 王念旭。DSP基础与应用系统设计。北京航空航天大学出版社，2000
[23] 刘海滨。数字信号处理系统及其应用。清华大学出版社，1997
[24] Texas Instruments Co. DSP Selection Guide. 2004
[25] 彭启琮，李玉柏，管庆。DSP技术的发展与应用。北京：高等教育出版社，2002
[26] Texas Instruments Co. TMS320C6414 DATASHEET. 2001
[27] Texas Instruments Co. Code Composer Studio Getting Started Guide, 2001
[28] Texas Instruments Co. TMS320C6000 Code Composer Studio Tutorial. 2000
[29] www.seeddsp.com
[30] 李方惠，任丽香等。TMS320C6000系列DSPs原理与应用。电子工业出版社，2003
[31] Texas Instruments Co. TMS320C6000 DSP Cache User's Guide. 2003
[32] Texas Instruments Co. TMS320C6000 DSP External Memory Interface (EMIF) Reference Guide. 2004
[33] Texas Instruments Co. TMS320C64x DSP Two-Level Internal Memory Reference Guide. 2002
[34] Texas Instruments Co. Using the TMS320C6000 McBSP as a High Speed Communication Port. 2001
[35] Texas Instruments Co. TMS320C6000 Technical Brief. 1999
[36] Texas Instruments Co. TMS320C6000 DSP Enhanced Direct Memory Access (EDMA) Controller Reference Guide. 2002
[37] Texas Instruments Co. TMS320C6000 DSP 32-bit Timer Reference Guide. 2001
[38] Texas Instruments Co. TMS320C6000 DSP Interrupt Selector Reference Guide. 2001
[39] Texas Instruments Co. TMS320C6000 Boot Mode and Emulation Reset. 2001
[40] Texas Instruments Co. TMS320C64x DSP Reference Design. 2003
[41] Texas Instruments Co. TMS320C6414/5/6 Power Consumption Summary. 2002
[42] Texas Instruments Co. 3.3V DSP for Digital Motor Control. 2000
[43] Texas Instruments Co. TPS54310 DATA SHEET. 2002
[44] Texas Instruments Co. TMS320C6000 Board Design for JTAG. 2000
[45] Texas Instruments Co. TMS320C6000 DSP Designing for JTAG Emulation Reference Guide.2000
[46] Texas Instruments Co. Emulation Fundamentals for Ti DSP Solutions.2000
[47] Texas Instruments Co. TMS320C6000 System Clock Circuit Example. 2001
[48] Integrated Circuit Systems, Inc. ICS512 DATASHEET.2002
[49] Texas Instruments Co. TMS320C6000 EMIF to External FIFO Interface.. 2002
[50] Cypress Semiconductor Corporation. CY7C421 DATASHEET.2002
[51] Texas Instruments Co. TMS320C6000 EMIF-to-External SDRAM Interface. 2003
[52] Texas Instruments Co. TMS320C6000 EMIF:Overview of Support of High Performance Memory Technology.2002
[53] Micron Technology, Inc. MT48LV4M32B2 DATASHEET. 2001
[54] Texas Instruments Co.TMS320C6000 EMIF to External Flash Memory.2002
[55] ADI.Am29LV800 DATASHEET. 2000
[56] ADI.Am29LV256 DATASHEET.2000
[57] http//:bbs.matwav.com..使用CCS进行DSP编程.
[58] Texas Instruments Co.Creating a Second-Level Bootloader for FLASH Bootloading on TMS320C6000 Platform With Code Composer Studio 2.2. 2004
[59] Texas Instruments Co. TMS320C6000 Programmer's Guide. 2000
[60] 张骏．基于DSP和CPLD的图像通信系统设计与实现．哈尔滨工程大学硕士学位论文，2004．2
[61] 黄正谨．CPLD系统设计技术入门与应用．电子工业出版社，2002
[62] ALTERA Co. ByteBlasterMV Parallel Port Download Cable. 1999
[63] http://www.edadoc.com/.PCB工艺规范
[64] VANTIS Co. High-Speed Board Design Techniques. 1997
[65] http://www.21ic.com/.印刷线路板设计指南
[66] http://www.cg163.net.Grounding of Mixed Signal PCBs
[67] Texas Instruments Co. Printed-Circuit Board Layout for Improved Electromagnetic Compatibility. 2003
[68] Texas Instruments Co. TMS320C6x Thermal Design Considerations. 1999
[69] Texas Instruments Co. TMS320C6000 BGA Manufacturing Considerations. 2001
[70] Robert Wang. Introduction To PCB Process Flow and Related Focusing Tech.北京英创力科技有限公司培训资料
[71] http://www.edadoc.com/.PCB载流能力计算．
[72] 康兵．ProtelDXP 2004应用与实例．国防工业出版社．2005
[73] 李珩，杨杉，欧大生．电路设计与制板：Protel DXP实用教程．西安电子科技大学出版社，2004
[74] 飞思科技产品研发中心 编著．神经网络理论与MATLAB7实现．电子工业出版社，2005
[75] 胡小锋，赵辉编著．Visual C++/MATLAB图像处理与识别实用案例精选．人民邮电出版社．2004
[76] Lalita SAKOOBUNTHU. Thai Banknote Recognition Using Neural Network.高知工科大学大学院基盤工科专攻情报工学，平成14年度修士学位论文
[77] 顾良．基于DSP自动指纹识别系统的研究与实现．电子科技大学硕士学位论文，2004