纸币号码识别系统的设计与研究
|
摘要
纸币号码的唯一性可以用来标识纸币的身份,利用嵌入式技术和数字图像识别技术对纸币号码识别系统进行研究,以实现纸币号码的自动识别和对金融流通领域的监管。本文从整体性的角度出发对纸币号码识别系统进行了研究,其主要内容如下:
1、硬件系统:通过对系统功能的综合考虑对下面几个部分进行了研究,包括图像采集模块,核心处理器,存储模块,通信模块,人机交互模块和电源模块等。并考虑了硬件抗干扰问题。
2、图像预处理:图像预处理是字符识别的基础,它是为后面的字符识别作准备。首先通过对纸币号码区特点的了解,利用先验知识实现对纸币号码区的快速定位;然后对粗定位的图像进行去噪、二值化、倾斜校正等工作;最后在精确定位的基础上实现了对字符的分割。
3、字符识别:基于结构特征分析判断法选取字符中相对稳定的特征生成判断树实现对字符的识别。随后基于对残字识别的考虑,对识别判断树进行了改进。
4、软件系统:对Linux系统进行裁剪、移植和引导并添加了相应设备的驱动,使系统得以支持这些设备的运行。
5、系统扩展功能:从实际应用和需求的角度出发,对系统的功能进行了扩展,引入了GUI图形界面,增加了对数据库的支持以及对系统升级的支持。
Paper currency numbers is unique, so it can be used to identify the paper money. Embedded technology and digital image recognition technology can be used to studythe paper money recognition system. The system either can be used to recognize the paper currency numbers automatically or can be used to supervise the financial circulation. The paper studies the system from an overall perspective. Its main contents are as follows:
1、The hardware system: From a comprehensive consideration, it contains the follows parts: image acquisition module, the core processor, memory module, communication module, human-computer interaction and power modules etc. It also considers the issue of hardware anti-jamming.
2、Image Pre-processing: Image Pre-processing is the basic of character recognition. First, we should understand the characteristics of the paper currency numbers. The priori knowledge is used to achieve the district's rapid positioning. Then, the following works are done for the rough positioning area, such as image de-noising, binarization, tilt correction, etc; Finally, we achieve segmention of characters based on the precise positioning.
3、Character recognition: The relatively stable features of characters are selected to make a identification tree. Based on a consideration of the residual words, the identification trees are improved.
4、Software systems: The following works are done for this part, such as cutting, grafting and guiding the linux system, adding the corresponding device driver, allowing the system to support the operation of the equipment.
5、Expanding the system functions: From the view of practical application, the GUI graphical interface is introduced. Finally we give the support for database and system upgrade.
1、硬件系统:通过对系统功能的综合考虑对下面几个部分进行了研究,包括图像采集模块,核心处理器,存储模块,通信模块,人机交互模块和电源模块等。并考虑了硬件抗干扰问题。
2、图像预处理:图像预处理是字符识别的基础,它是为后面的字符识别作准备。首先通过对纸币号码区特点的了解,利用先验知识实现对纸币号码区的快速定位;然后对粗定位的图像进行去噪、二值化、倾斜校正等工作;最后在精确定位的基础上实现了对字符的分割。
3、字符识别:基于结构特征分析判断法选取字符中相对稳定的特征生成判断树实现对字符的识别。随后基于对残字识别的考虑,对识别判断树进行了改进。
4、软件系统:对Linux系统进行裁剪、移植和引导并添加了相应设备的驱动,使系统得以支持这些设备的运行。
5、系统扩展功能:从实际应用和需求的角度出发,对系统的功能进行了扩展,引入了GUI图形界面,增加了对数据库的支持以及对系统升级的支持。
Paper currency numbers is unique, so it can be used to identify the paper money. Embedded technology and digital image recognition technology can be used to studythe paper money recognition system. The system either can be used to recognize the paper currency numbers automatically or can be used to supervise the financial circulation. The paper studies the system from an overall perspective. Its main contents are as follows:
1、The hardware system: From a comprehensive consideration, it contains the follows parts: image acquisition module, the core processor, memory module, communication module, human-computer interaction and power modules etc. It also considers the issue of hardware anti-jamming.
2、Image Pre-processing: Image Pre-processing is the basic of character recognition. First, we should understand the characteristics of the paper currency numbers. The priori knowledge is used to achieve the district's rapid positioning. Then, the following works are done for the rough positioning area, such as image de-noising, binarization, tilt correction, etc; Finally, we achieve segmention of characters based on the precise positioning.
3、Character recognition: The relatively stable features of characters are selected to make a identification tree. Based on a consideration of the residual words, the identification trees are improved.
4、Software systems: The following works are done for this part, such as cutting, grafting and guiding the linux system, adding the corresponding device driver, allowing the system to support the operation of the equipment.
5、Expanding the system functions: From the view of practical application, the GUI graphical interface is introduced. Finally we give the support for database and system upgrade.
引文
[1]周翟和.纸币读入识别系统的研究和实现[D].南京航空航天大学硕士论文.2002.
[2]余国华,冯启明.基于COMS图像传感器的视频采集系统设计[J].武汉理工大学学报.2004,No1,Vol.28:145-147.
[3]宋勇,郝群等.基于CMOS图像传感器的成像系统的设计[J].光学技术.2002,No3.Vol.28:253-256.
[4]OV7620 Data Sheet.OmniVision Inc.2001.
[5]刘建业.纸币号码读入识别系统的单片机实现[J].工业控制计算机.2002,No.1,Vol.15:69-71.
[6]连华,林斌等.COMS图像传感器成像系统[J].光学仪器,2003,No.4,Vol.25:12-19.
[7]田泽编著.ARM9嵌入式开发实验与实践[M].北京:北京航空大学出版社,2006.
[8]ARM9TDMI Data Sheet.ARM Ltd.1995.
[9]ARM9基础实验教程.深圳英蓓特信息技术有限公司.2005.
[10]HY57V561620 Data Sheet.Hynix Semiconductor Inc.1998.
[11]CS8900A Data Sheet.Cirrus logic Inc.2001.
[12]mini2440用户手册.广州友善之臂计算机科技有限公司.2009.
[13]孙天泽,袁文菊,张海峰等编著.嵌入式设计及Linux驱动开发指南[M].北京:电子工业出版社,2005.
[14]http://zhidao.baidu.com.
[15]裴志军,国澄明等.CMOS有源像素图像传感器的噪声控制技术[J].半导体光电.2002,No.6,Vol.23:382-385.
[16]徐慧等编著.Visual C++数字图像实用工程案例精选[M].北京:人民邮电出版社,2004.
[17]E Abreu and S.K.Mirtra.A signal-dependent rank ordered mean(SD-ROM)filter-a new approach for removal of impulses from highly corrupted images[M].IEEE Image Processing.1995,Vol.5:2371-2374.
[18]How-Lung End and Kai-Kung Ma.Nosie Adaptive Soft-Switching Median Filter[J].IEEE Trans on Image Processing.2001,No2,Vol.10:242-258.
[19]王德辉,张世林等.CCD技术与图像测量在叶片测量中的应用[J].计量与测试技术.2002,No.2:17-21.
[20]荣观澳编著.计算机图像处理[M].北京:清华大学出版社.2000.
[21]Jain A K,Yu B.Automatic Text Location in Images and Video Frames[M].Pattern Recognition.1998,No12,Vol.31:2055-2076.
[22]Nikhil R P,Sankar K P.A Review on Image Segmentation Techniques[J].Pattern Recognition.1993,No.9,Vol.26:1277-1294.
[23]Rafael C.Gonzalez,Richard,E.Woods.Digital Image Processing[M].Beijing:china Machine Press.2003,Second Edition.
[24]吴辰铌.基于DSP号码识别系统及实现[D].哈尔滨工业大学硕士论文.2004.
[25]Vincent L,Soille P.Watersheds in digital spaces:an efficient algorithm based on immersion simulations[J].IEEE Transactions on Pattern Analysis and Machine Intelligence.1991,No.6,Vol.13:583-598.
[26]叶青,张春华.汽车牌照实时几何畸变校正方法[J].红外与激光工程.2004,No.2,Vol.33:181-184.
[27]余小宝.增值税发票抵扣联字符识别中的图像倾斜校正方法[J].电脑编程技巧与维护.2004,No.3:14-16.
[28]周海林,王立琦.光学图像几何畸变的快速校正算法[J].中国图像图形学报.2003,No.10,Vol.8:1131-1335.
[29]李迎春,曾峦等.图像的几何运算方法[J].指挥技术学院学报.2000,No.1,Vol.11:46-49.
[30]郑南宁.计算机视觉与模式识别[M].北京:国防工业出版社.1998.
[31]范勇,蒋欣荣等.汽车牌照快速定位算法[J].光电工程学报.2001,No.2,Vol.28:56-59.
[32]李宏升,马洪庆,刘济林等.利用牌照识别技术的停车场安全防盗系统[J].计算机系统应用.1999,No.5:14-16.
[33]吕雪峰,王敏,万国红.汽车牌照定位技术研究[J].计算机工程与应用.2003,No.15:226-229.
[34]刘伟铭,赵雪平.一种基于扫描行的汽车车牌定位算法[J].计算机工程与应用.2004.No6:223-225.
[35]Mori S,Suen C Y,Yamamoto K.Historical Review of OCR Research and Development[J].Proc of IEEE TransPAMI.1996,No.7,Vol.18:690-706.
[36]荆涛,王中.光学字符识别技术与展望[J].计算机工程.2003,No.2,Vol.29:1-2.
[37]杜彦蕊,杜海明.手写字符识别方法的研究[J].应用科技.2004,No.2,Vol.31:19-21.
[38]刘刚,张洪刚,郭军.用于脱机手写数字识别的隐马尔科夫模型[J].计算机研究与发展.2003,No.8,Vol.40:1252-1257.
[39]高彦宇,杨杨.脱机手写汉字识别研究综述[J].计算机工程与应用.2004,No.7: 74-77.
[40]Nei kato,Shin chiro Omachi.A handwriting character recognition system using direction element feature[J].IEEE Trans on Pattern Analysis and Machine Intellligence.1999,No.3,Vol.21:258-262.
[41]罗勇,周必水.一种基于参照模型的联机手写汉字识别方法[J].杭州电子工业学院学报.2003,No.3,Vol.23:66-69.
[42]鲁湛,丁晓青.基于笔段间关系的联机手写汉字HMM模型[J].清华大学学报(自然科学版).2004,No.7,Vol.44:913-16.
[43]Takahashi K,Yasuda H,Matsumoto T.A fast HMM algorithm for on-line handwritten character recognition[C].Document Analysis and Recognition.1997,Vol.1:369-375.
[44]赵巍,刘家锋,唐降龙.基于部件HMM级联的联机手写汉字识别方法[J].哈尔滨工业大学学报.2004,No.5,Vol.36:570-573.
[45]Oiviand Due Trie,Anil K Jain,Torfinn Taxt.Feature Extraction Methods for Character Recognition-A Survey[J].Pattern Recognition.1996,No.4,Vol.29:641-662.
[46]郭晓松,孙祥玉,扬必武.基于连通域的模板匹配法用于字符识别的算法研究[J].计算机工程与应用.2000,No.12:46-47.
[47]阐伟,朱秋煌.一种改进的基于模板匹配的集装箱字符识别方法的研究[J].计算机工程.2000,No.12,Vol.26:119-121.
[48]高彤,姜华,吕民.基于模板匹配的手写体字符识别方法[J].哈尔滨工业大学学报.1999,No.1,Vol.31:104-106.
[49]陈斌,游志胜.车牌号码松弛模板匹配方法[J].计算机应用.2001,No.8,Vol.21:155-157.
[50]陈芳信,马翔,姜新等.基于多神经网络的手写数字综合识别系统[J].信息技术.2003,No.9,Vol.27:1-4.
[51]孙立民,张宗念,余英林等.有监督竞争学习神经模型在手写体汉字识别中的应用[J].工程图学报.1998,No.4:75-80.
[52]吴鸣锐,张钹.一种用于大规模模式识别问题的神经网络算法[J].软件学报.2001,No.6,Vol.21:851-855.
[53]扬万山,陈松桥,唐连章.基于BP神经网络的工程图纸图形符号的识别[J].微型电脑应用.2000,No.2,Vol.16:22-23.
[54]闫巧,王世军,谢维新等.基于遗传--神经网络的字符识别[J].兰州铁道学院学报(自然科学版).2001,No.4,Vol.20:78-81.
[55]张丰收,崔凤奎,王晓强等.BP网络在工程图自动识别中的应用[J].拖拉机与农用运输车.2000,No.4:43-45.
[56]武强,童雪锋,季隽.基于人工神经网络的数字字符识别[J].计算机工程.2003,No.14,Vol.29:112-114.
[57]杨勇,谢刚生.基于BP神经网络的手写数字识别[J].华东地质学院学报.2003,No.4,Vol.26:383-386.
[58]T.K.Ho,etc.Decision Combination in Multiple Classifier Systems[J].IEEE Trans On Pattern Analysis and Machine Intelligence.1994,No.1,Vol.16:145-154.
[59]L.Xu,etc.Methods of Combining Multiple Classifiers and Their Applications to Hand-written Recognition[J].IEEE Trans.on System,Man,and Cybernetics.1992,No.3,Vol.22:418-435.
[60]宋建才.字符结构知识在车牌识别中的应用[J].计算机应用.2004,No.4:18-19.
[61]王钿.卓兴旺编著.基于Verilog HDL的数字系统应用设计[M].北京:国防工业出版社.2006.
[62]李善平,刘文峰,王焕龙.Linux与嵌入式系统[M].北京:清华大学出版社.2000.
[63]詹昌平.基于嵌入式Linux的货币识别接收器设计与实现[D].中南大学硕士论文.2005.
[2]余国华,冯启明.基于COMS图像传感器的视频采集系统设计[J].武汉理工大学学报.2004,No1,Vol.28:145-147.
[3]宋勇,郝群等.基于CMOS图像传感器的成像系统的设计[J].光学技术.2002,No3.Vol.28:253-256.
[4]OV7620 Data Sheet.OmniVision Inc.2001.
[5]刘建业.纸币号码读入识别系统的单片机实现[J].工业控制计算机.2002,No.1,Vol.15:69-71.
[6]连华,林斌等.COMS图像传感器成像系统[J].光学仪器,2003,No.4,Vol.25:12-19.
[7]田泽编著.ARM9嵌入式开发实验与实践[M].北京:北京航空大学出版社,2006.
[8]ARM9TDMI Data Sheet.ARM Ltd.1995.
[9]ARM9基础实验教程.深圳英蓓特信息技术有限公司.2005.
[10]HY57V561620 Data Sheet.Hynix Semiconductor Inc.1998.
[11]CS8900A Data Sheet.Cirrus logic Inc.2001.
[12]mini2440用户手册.广州友善之臂计算机科技有限公司.2009.
[13]孙天泽,袁文菊,张海峰等编著.嵌入式设计及Linux驱动开发指南[M].北京:电子工业出版社,2005.
[14]http://zhidao.baidu.com.
[15]裴志军,国澄明等.CMOS有源像素图像传感器的噪声控制技术[J].半导体光电.2002,No.6,Vol.23:382-385.
[16]徐慧等编著.Visual C++数字图像实用工程案例精选[M].北京:人民邮电出版社,2004.
[17]E Abreu and S.K.Mirtra.A signal-dependent rank ordered mean(SD-ROM)filter-a new approach for removal of impulses from highly corrupted images[M].IEEE Image Processing.1995,Vol.5:2371-2374.
[18]How-Lung End and Kai-Kung Ma.Nosie Adaptive Soft-Switching Median Filter[J].IEEE Trans on Image Processing.2001,No2,Vol.10:242-258.
[19]王德辉,张世林等.CCD技术与图像测量在叶片测量中的应用[J].计量与测试技术.2002,No.2:17-21.
[20]荣观澳编著.计算机图像处理[M].北京:清华大学出版社.2000.
[21]Jain A K,Yu B.Automatic Text Location in Images and Video Frames[M].Pattern Recognition.1998,No12,Vol.31:2055-2076.
[22]Nikhil R P,Sankar K P.A Review on Image Segmentation Techniques[J].Pattern Recognition.1993,No.9,Vol.26:1277-1294.
[23]Rafael C.Gonzalez,Richard,E.Woods.Digital Image Processing[M].Beijing:china Machine Press.2003,Second Edition.
[24]吴辰铌.基于DSP号码识别系统及实现[D].哈尔滨工业大学硕士论文.2004.
[25]Vincent L,Soille P.Watersheds in digital spaces:an efficient algorithm based on immersion simulations[J].IEEE Transactions on Pattern Analysis and Machine Intelligence.1991,No.6,Vol.13:583-598.
[26]叶青,张春华.汽车牌照实时几何畸变校正方法[J].红外与激光工程.2004,No.2,Vol.33:181-184.
[27]余小宝.增值税发票抵扣联字符识别中的图像倾斜校正方法[J].电脑编程技巧与维护.2004,No.3:14-16.
[28]周海林,王立琦.光学图像几何畸变的快速校正算法[J].中国图像图形学报.2003,No.10,Vol.8:1131-1335.
[29]李迎春,曾峦等.图像的几何运算方法[J].指挥技术学院学报.2000,No.1,Vol.11:46-49.
[30]郑南宁.计算机视觉与模式识别[M].北京:国防工业出版社.1998.
[31]范勇,蒋欣荣等.汽车牌照快速定位算法[J].光电工程学报.2001,No.2,Vol.28:56-59.
[32]李宏升,马洪庆,刘济林等.利用牌照识别技术的停车场安全防盗系统[J].计算机系统应用.1999,No.5:14-16.
[33]吕雪峰,王敏,万国红.汽车牌照定位技术研究[J].计算机工程与应用.2003,No.15:226-229.
[34]刘伟铭,赵雪平.一种基于扫描行的汽车车牌定位算法[J].计算机工程与应用.2004.No6:223-225.
[35]Mori S,Suen C Y,Yamamoto K.Historical Review of OCR Research and Development[J].Proc of IEEE TransPAMI.1996,No.7,Vol.18:690-706.
[36]荆涛,王中.光学字符识别技术与展望[J].计算机工程.2003,No.2,Vol.29:1-2.
[37]杜彦蕊,杜海明.手写字符识别方法的研究[J].应用科技.2004,No.2,Vol.31:19-21.
[38]刘刚,张洪刚,郭军.用于脱机手写数字识别的隐马尔科夫模型[J].计算机研究与发展.2003,No.8,Vol.40:1252-1257.
[39]高彦宇,杨杨.脱机手写汉字识别研究综述[J].计算机工程与应用.2004,No.7: 74-77.
[40]Nei kato,Shin chiro Omachi.A handwriting character recognition system using direction element feature[J].IEEE Trans on Pattern Analysis and Machine Intellligence.1999,No.3,Vol.21:258-262.
[41]罗勇,周必水.一种基于参照模型的联机手写汉字识别方法[J].杭州电子工业学院学报.2003,No.3,Vol.23:66-69.
[42]鲁湛,丁晓青.基于笔段间关系的联机手写汉字HMM模型[J].清华大学学报(自然科学版).2004,No.7,Vol.44:913-16.
[43]Takahashi K,Yasuda H,Matsumoto T.A fast HMM algorithm for on-line handwritten character recognition[C].Document Analysis and Recognition.1997,Vol.1:369-375.
[44]赵巍,刘家锋,唐降龙.基于部件HMM级联的联机手写汉字识别方法[J].哈尔滨工业大学学报.2004,No.5,Vol.36:570-573.
[45]Oiviand Due Trie,Anil K Jain,Torfinn Taxt.Feature Extraction Methods for Character Recognition-A Survey[J].Pattern Recognition.1996,No.4,Vol.29:641-662.
[46]郭晓松,孙祥玉,扬必武.基于连通域的模板匹配法用于字符识别的算法研究[J].计算机工程与应用.2000,No.12:46-47.
[47]阐伟,朱秋煌.一种改进的基于模板匹配的集装箱字符识别方法的研究[J].计算机工程.2000,No.12,Vol.26:119-121.
[48]高彤,姜华,吕民.基于模板匹配的手写体字符识别方法[J].哈尔滨工业大学学报.1999,No.1,Vol.31:104-106.
[49]陈斌,游志胜.车牌号码松弛模板匹配方法[J].计算机应用.2001,No.8,Vol.21:155-157.
[50]陈芳信,马翔,姜新等.基于多神经网络的手写数字综合识别系统[J].信息技术.2003,No.9,Vol.27:1-4.
[51]孙立民,张宗念,余英林等.有监督竞争学习神经模型在手写体汉字识别中的应用[J].工程图学报.1998,No.4:75-80.
[52]吴鸣锐,张钹.一种用于大规模模式识别问题的神经网络算法[J].软件学报.2001,No.6,Vol.21:851-855.
[53]扬万山,陈松桥,唐连章.基于BP神经网络的工程图纸图形符号的识别[J].微型电脑应用.2000,No.2,Vol.16:22-23.
[54]闫巧,王世军,谢维新等.基于遗传--神经网络的字符识别[J].兰州铁道学院学报(自然科学版).2001,No.4,Vol.20:78-81.
[55]张丰收,崔凤奎,王晓强等.BP网络在工程图自动识别中的应用[J].拖拉机与农用运输车.2000,No.4:43-45.
[56]武强,童雪锋,季隽.基于人工神经网络的数字字符识别[J].计算机工程.2003,No.14,Vol.29:112-114.
[57]杨勇,谢刚生.基于BP神经网络的手写数字识别[J].华东地质学院学报.2003,No.4,Vol.26:383-386.
[58]T.K.Ho,etc.Decision Combination in Multiple Classifier Systems[J].IEEE Trans On Pattern Analysis and Machine Intelligence.1994,No.1,Vol.16:145-154.
[59]L.Xu,etc.Methods of Combining Multiple Classifiers and Their Applications to Hand-written Recognition[J].IEEE Trans.on System,Man,and Cybernetics.1992,No.3,Vol.22:418-435.
[60]宋建才.字符结构知识在车牌识别中的应用[J].计算机应用.2004,No.4:18-19.
[61]王钿.卓兴旺编著.基于Verilog HDL的数字系统应用设计[M].北京:国防工业出版社.2006.
[62]李善平,刘文峰,王焕龙.Linux与嵌入式系统[M].北京:清华大学出版社.2000.
[63]詹昌平.基于嵌入式Linux的货币识别接收器设计与实现[D].中南大学硕士论文.2005.