基于DSP的车牌识别系统研究与设计
摘要
智能交通系统(ITS)是未来交通发展的必然趋势,是目前电子信息技术在交通运输领域应用的前沿研究课题。车牌识别系统(License Plate Recognition System,LPRS)是智能交通系统的重要组成部分,在规范交通管理及减少交通事故等方面具有重要的现实意义。近年来,人们对车牌自动识别技术开展了广泛的研究,但由于算法的复杂性以及系统实现的困难性,目前仍存在一些急待解决的问题,有必要对这一技术进行深入的研究。
本文对车牌识别过程中的一些难点进行了研究探讨,阐述了一套车牌识别的算法体系。在车牌定位以及倾斜校正、字符分割方面给出了自己的研究成果。提出了一种综合车牌纹理、形状、颜色特征的定位方法和基于Radon变换和模板匹配的字符分割方法,最后论述了一种基于改进BP算法的字符识别方法。
在应用上以本研究室交通参数检测方面的研究成果为基础,以DSP处理器为核心,通过采用DSP/BIOS实时操作系统,利用基于图像处理的计算机视觉技术,设计出分布式处理车牌实时识别系统。系统通过交通流检测、车牌识别、图像压缩与数据上载等模块,实现车牌实时识别。
在软件设计上采用了基于类/微驱动(class/mini-driver)模型的开发方式。类/微驱动是在DSP/BIOS上构建的一种分层驱动模型,由类驱动和微驱动两部分构成。本文对DSP图像采集模块和HPI接口模块开发出微驱动,将算法实现与硬件平台独立开来。有效的提高了开发效率和软件复用性。
Intelligent Transport System (ITS) is the development trend in the future. It is the leading research topic of current electronic information technology applied in the realm of transportation. Be an important part of ITS, the License Plate Recognition System(LPRS) is of much significance in regulating the transportation and reducing the cases of traffic accident. The License Plate Recognition(LPR) is widely researched recently, while due to the complexity of the algorithm and difficulty of realization, several urgent puzzles remain unraveled, therefore more research should be paid to this technique.
This thesis surveys certain difficult points in the process of LPR, and expounds a series of algorithms of it. It presents my accomplishment in License-plate location, Skew Correction and segmentation of Vehicle License Plate. An edge-texture-color based algorithm to locate the plate region and segmentation algorithm based on template matching and Radon transform are proposed. At last it maintains an enhanced BP algorithm in character recognition.
With the basis of the achievement in transportation parameter measurement of our lab, this thesis gives a distributed LPRS based on DSP by using the DSP/BIOS and utilizing the computer to deal with the picture, to imitate people’s vision function. The system consists of industrial control computer and DSP real-time sampling cards, by means of video capture, image processing, image compressing and data upload, it can perform Real-time LPR.
It adopts a new development mode based on class/mini-driver model in the software development. The driver is a Two-Level Device Model which composed by Class driver and mini-driver. According to this model the mini-driver of Image Acquisition Module and HPI Module are developed. These separate the algorithms development and hardware platform, and boost the efficiency of development and the reusable of algorithms.
本文对车牌识别过程中的一些难点进行了研究探讨,阐述了一套车牌识别的算法体系。在车牌定位以及倾斜校正、字符分割方面给出了自己的研究成果。提出了一种综合车牌纹理、形状、颜色特征的定位方法和基于Radon变换和模板匹配的字符分割方法,最后论述了一种基于改进BP算法的字符识别方法。
在应用上以本研究室交通参数检测方面的研究成果为基础,以DSP处理器为核心,通过采用DSP/BIOS实时操作系统,利用基于图像处理的计算机视觉技术,设计出分布式处理车牌实时识别系统。系统通过交通流检测、车牌识别、图像压缩与数据上载等模块,实现车牌实时识别。
在软件设计上采用了基于类/微驱动(class/mini-driver)模型的开发方式。类/微驱动是在DSP/BIOS上构建的一种分层驱动模型,由类驱动和微驱动两部分构成。本文对DSP图像采集模块和HPI接口模块开发出微驱动,将算法实现与硬件平台独立开来。有效的提高了开发效率和软件复用性。
Intelligent Transport System (ITS) is the development trend in the future. It is the leading research topic of current electronic information technology applied in the realm of transportation. Be an important part of ITS, the License Plate Recognition System(LPRS) is of much significance in regulating the transportation and reducing the cases of traffic accident. The License Plate Recognition(LPR) is widely researched recently, while due to the complexity of the algorithm and difficulty of realization, several urgent puzzles remain unraveled, therefore more research should be paid to this technique.
This thesis surveys certain difficult points in the process of LPR, and expounds a series of algorithms of it. It presents my accomplishment in License-plate location, Skew Correction and segmentation of Vehicle License Plate. An edge-texture-color based algorithm to locate the plate region and segmentation algorithm based on template matching and Radon transform are proposed. At last it maintains an enhanced BP algorithm in character recognition.
With the basis of the achievement in transportation parameter measurement of our lab, this thesis gives a distributed LPRS based on DSP by using the DSP/BIOS and utilizing the computer to deal with the picture, to imitate people’s vision function. The system consists of industrial control computer and DSP real-time sampling cards, by means of video capture, image processing, image compressing and data upload, it can perform Real-time LPR.
It adopts a new development mode based on class/mini-driver model in the software development. The driver is a Two-Level Device Model which composed by Class driver and mini-driver. According to this model the mini-driver of Image Acquisition Module and HPI Module are developed. These separate the algorithms development and hardware platform, and boost the efficiency of development and the reusable of algorithms.
引文
[1]朱东辉.智能交通系统的发展.山东交通学院学报, 2002, 10(4):9-15
[2]李彦旭,成立. ITS在国内外的发展趋势及我国的对策综述.光电工程, 2001, 31(12):40-45.
[3]夏劲,郭红卫.国内外城市智能交通系统的发展概况与趋势及其启示.科技进步与对策, 2003, 1:176-179
[4] Roy.L, Courtney. A broad view of ITS standards in the US. Intelligent Transportation System, 1997,11:529-536
[5]刘允才.智能交通国际发展概况和国内优先考虑的课题.公路, 2001, 11: 48-53
[6] RAFAEL,M. INIGO. Traffic monitoring and control using machine vision: a survey.IEEE TRANS on industrial electronics, 1995. 8, 32(8): 177-185
[7]张树波,赖剑煌.车牌定位和分割的一种综合方法.中山大学学报(自然科学版), 2004, 43(2):276-278
[8]刘治.实时自动车牌识别系统的研究:硕士学位论文.山东大学图书馆, 2004.
[9]胡松利,张斌,付江成等.具有车牌车型自动识别功能的路桥收费系统研究.交通与计算机, 2006, 24(3):34-35
[10]李方慧,王飞,何佩琨. TMS320C6000系列DSPs原理与应用(第二版).北京:电子工业出版社, 2003: 24-30
[11]许伟. DSP应用的结构和发展方向.计算机应用, 1999, 3:20-22.
[12]张绍满.基于DSP分布式处理的视频检测系统研究与设计.硕士学位论文,华中科技大学图书馆, 2004
[13]TMS320C6000 EMIF-to-External SDRAM Interface. Texas Instruments Incorp- orated , December 2001
[14]TMS320C6000 HPI To PCI Interfacing Using The PLX PCI9050. Texas Instruments Incorporated , April 2002
[15]Yasuharu Yanamura, Masahiro Goto, Daisuke Nishiyama. Extraction and Tracking of the License Plate Using Hough Transform and Voted Block Matching.IEEE Intelligent Vehicles Symposium Conference,2003,6:243- 246
[16]张引,潘云鹤.彩色汽车图像牌照定位新方法.中国图像图形学报, 2001, 6A(4): 374-377
[17]M.H.ter, Brugge, J.h.Stevens, J.A.G.Nijhuis. License Plate Recognition Using DTCN- Ns. Proceeding of the IEEE International Workshop on Cellular Neural NetWorks and their Application,1998:212-217
[18]PARISIR.Car Plate Recognition by Neural Networks and Image Processing. Circuits and Systems, 1998, 3(5):195-198.
[19]管庶安.按综合特征实现车牌快速定位.计算机应用与软件, 2005, 22 (6):63-64
[20]陈寅鹏,丁晓青.复杂车辆图像中的车牌定位与字符分割方法.红外与激光工程. 2004, 33(1): 29-33
[21]郭大波.基于车牌底色识别的车牌定位方法.计算机工程与设计, 2003, 24(5): 81-87
[22]Wen C-Y,Yu C-C,Hun Z-D.A 3-D transformation to improve the legibility of license plate numbers. Journal of Forensic Sciences,2002,47(3):578-585
[23]郝永杰,刘文耀,路烁.畸变汽车牌照图像的空间校正.西南交通大学学报, 2002, 37(4): 417-420
[24]王金平.平面Radon变换的反演. CT理论与应用研究, 2000, 9(1): 8-11
[25]何斌,马天予,王运坚等. Visual C++数字图像处理(第一版).北京:人民邮电出版社, 2001: 199-206
[26]胡小锋,赵辉. Visual C++/MATLAB图像处理与识别.北京:人民邮电出版社, 2004:135-152
[27]范玮琦,穆长江.一种基于汉字结构特征的车牌照字符分割方法.仪器仪表学报, 2003, 24(4):472-474
[28]Hu Aiming, Zhou Xiaokuan. Image segmentation of carlicense plate using morpho- logic features. Journal of Computer-Aided Design&Computer Graphics,2003, 15(6): 716-719
[29]刘明军,谢宏霖,孙雪松.车牌字符分割算法的比较研究.济南大学学报(自然科学版). 2006, 6(3): 246-248
[30]陈寅鹏,丁晓青.复杂车辆图像中的车牌定位与字符分割方法.红外与激光工程, 2004, 2(1):29-33
[31]宋加涛,刘济林.车辆牌照上英文和数字字符的结构特征分析及提取.中国图像图形学报, 2002, 7A(9):945-949.
[32]宋建才.字符结构知识在车牌识别中的应用.电子技术应用, 2004, 4:18-19
[33]Parist R. Car plate crcognition by neural nerworks and image processing. IEEE International Symposium on Circuits and Systems, USA. 2000, 5(3): 31-37
[34]G.Nijhuis.Car License Plate Recognition with Neural Networks and Fuzzy Logic. Proceedings 1998 Fifth IEEE International Workshop on Publish,1998:212-217.
[35]Park S H, Kim K I, Jung K, etal. Locating car license plate using neural networks. Electronics Letters,1999, 35(17):1475-1477
[36]谭三,盛翊智.自适应斜率输出函数的BP算法.华中科技大学学报, 2001, 29(12): 31-33
[37]TMS320 DSP/BIOS User's Guide. Texas Instruments Incorporated. Texas Instruments Incorporated ,November 2002
[38]TMS320C6000 Chip Support Library API Reference Guide. Texas InstrumentsIncorporated December 2002
[39]DSP/BIOS Driver Developer's Guide. Texas Instruments Incorporated, November 2002.
[40]TMS320C6000 DSP/BIOS Application Programming Interface Reference Guide. Texas Instruments Incorporated. Texas Instruments Incorporated , November 2002
[41]The TMS320DM642 Video Port Mini2Driver. Texas Instruments Incorporated , August,2003
[42]Image Processing Examples Using the TMS320C62xImage/Video Processing Library. Texas Instruments Incorporated, March 2003
[43]A DSP/BIOS PCI Device Driver for TMS320C64xx DSPs . Texas Instruments Incorporated, June 2003
[44]耿倩,王耀南,余浩.基于DSP的JPEG静态图像压缩编码设计与实现.计算机工程与应用, 2006, (23): 82-84
[45]曾国良,丁宗豪.静像数据压缩及其标准JPEG.电视技术, 1994, 4:2-5
[46]TMS320C6x Optimizing C Compiler. Texas Instruments Incorporated, February 1998
[47]Optimizing JPEG on the TMS320C6211 2-Level Cache. Texas Instruments Incorporated, December 2000
[48]TMS320 DSP Algorithm Standard Rules and Guidelines. Texas Instruments Incorporated, 2001
[2]李彦旭,成立. ITS在国内外的发展趋势及我国的对策综述.光电工程, 2001, 31(12):40-45.
[3]夏劲,郭红卫.国内外城市智能交通系统的发展概况与趋势及其启示.科技进步与对策, 2003, 1:176-179
[4] Roy.L, Courtney. A broad view of ITS standards in the US. Intelligent Transportation System, 1997,11:529-536
[5]刘允才.智能交通国际发展概况和国内优先考虑的课题.公路, 2001, 11: 48-53
[6] RAFAEL,M. INIGO. Traffic monitoring and control using machine vision: a survey.IEEE TRANS on industrial electronics, 1995. 8, 32(8): 177-185
[7]张树波,赖剑煌.车牌定位和分割的一种综合方法.中山大学学报(自然科学版), 2004, 43(2):276-278
[8]刘治.实时自动车牌识别系统的研究:硕士学位论文.山东大学图书馆, 2004.
[9]胡松利,张斌,付江成等.具有车牌车型自动识别功能的路桥收费系统研究.交通与计算机, 2006, 24(3):34-35
[10]李方慧,王飞,何佩琨. TMS320C6000系列DSPs原理与应用(第二版).北京:电子工业出版社, 2003: 24-30
[11]许伟. DSP应用的结构和发展方向.计算机应用, 1999, 3:20-22.
[12]张绍满.基于DSP分布式处理的视频检测系统研究与设计.硕士学位论文,华中科技大学图书馆, 2004
[13]TMS320C6000 EMIF-to-External SDRAM Interface. Texas Instruments Incorp- orated , December 2001
[14]TMS320C6000 HPI To PCI Interfacing Using The PLX PCI9050. Texas Instruments Incorporated , April 2002
[15]Yasuharu Yanamura, Masahiro Goto, Daisuke Nishiyama. Extraction and Tracking of the License Plate Using Hough Transform and Voted Block Matching.IEEE Intelligent Vehicles Symposium Conference,2003,6:243- 246
[16]张引,潘云鹤.彩色汽车图像牌照定位新方法.中国图像图形学报, 2001, 6A(4): 374-377
[17]M.H.ter, Brugge, J.h.Stevens, J.A.G.Nijhuis. License Plate Recognition Using DTCN- Ns. Proceeding of the IEEE International Workshop on Cellular Neural NetWorks and their Application,1998:212-217
[18]PARISIR.Car Plate Recognition by Neural Networks and Image Processing. Circuits and Systems, 1998, 3(5):195-198.
[19]管庶安.按综合特征实现车牌快速定位.计算机应用与软件, 2005, 22 (6):63-64
[20]陈寅鹏,丁晓青.复杂车辆图像中的车牌定位与字符分割方法.红外与激光工程. 2004, 33(1): 29-33
[21]郭大波.基于车牌底色识别的车牌定位方法.计算机工程与设计, 2003, 24(5): 81-87
[22]Wen C-Y,Yu C-C,Hun Z-D.A 3-D transformation to improve the legibility of license plate numbers. Journal of Forensic Sciences,2002,47(3):578-585
[23]郝永杰,刘文耀,路烁.畸变汽车牌照图像的空间校正.西南交通大学学报, 2002, 37(4): 417-420
[24]王金平.平面Radon变换的反演. CT理论与应用研究, 2000, 9(1): 8-11
[25]何斌,马天予,王运坚等. Visual C++数字图像处理(第一版).北京:人民邮电出版社, 2001: 199-206
[26]胡小锋,赵辉. Visual C++/MATLAB图像处理与识别.北京:人民邮电出版社, 2004:135-152
[27]范玮琦,穆长江.一种基于汉字结构特征的车牌照字符分割方法.仪器仪表学报, 2003, 24(4):472-474
[28]Hu Aiming, Zhou Xiaokuan. Image segmentation of carlicense plate using morpho- logic features. Journal of Computer-Aided Design&Computer Graphics,2003, 15(6): 716-719
[29]刘明军,谢宏霖,孙雪松.车牌字符分割算法的比较研究.济南大学学报(自然科学版). 2006, 6(3): 246-248
[30]陈寅鹏,丁晓青.复杂车辆图像中的车牌定位与字符分割方法.红外与激光工程, 2004, 2(1):29-33
[31]宋加涛,刘济林.车辆牌照上英文和数字字符的结构特征分析及提取.中国图像图形学报, 2002, 7A(9):945-949.
[32]宋建才.字符结构知识在车牌识别中的应用.电子技术应用, 2004, 4:18-19
[33]Parist R. Car plate crcognition by neural nerworks and image processing. IEEE International Symposium on Circuits and Systems, USA. 2000, 5(3): 31-37
[34]G.Nijhuis.Car License Plate Recognition with Neural Networks and Fuzzy Logic. Proceedings 1998 Fifth IEEE International Workshop on Publish,1998:212-217.
[35]Park S H, Kim K I, Jung K, etal. Locating car license plate using neural networks. Electronics Letters,1999, 35(17):1475-1477
[36]谭三,盛翊智.自适应斜率输出函数的BP算法.华中科技大学学报, 2001, 29(12): 31-33
[37]TMS320 DSP/BIOS User's Guide. Texas Instruments Incorporated. Texas Instruments Incorporated ,November 2002
[38]TMS320C6000 Chip Support Library API Reference Guide. Texas InstrumentsIncorporated December 2002
[39]DSP/BIOS Driver Developer's Guide. Texas Instruments Incorporated, November 2002.
[40]TMS320C6000 DSP/BIOS Application Programming Interface Reference Guide. Texas Instruments Incorporated. Texas Instruments Incorporated , November 2002
[41]The TMS320DM642 Video Port Mini2Driver. Texas Instruments Incorporated , August,2003
[42]Image Processing Examples Using the TMS320C62xImage/Video Processing Library. Texas Instruments Incorporated, March 2003
[43]A DSP/BIOS PCI Device Driver for TMS320C64xx DSPs . Texas Instruments Incorporated, June 2003
[44]耿倩,王耀南,余浩.基于DSP的JPEG静态图像压缩编码设计与实现.计算机工程与应用, 2006, (23): 82-84
[45]曾国良,丁宗豪.静像数据压缩及其标准JPEG.电视技术, 1994, 4:2-5
[46]TMS320C6x Optimizing C Compiler. Texas Instruments Incorporated, February 1998
[47]Optimizing JPEG on the TMS320C6211 2-Level Cache. Texas Instruments Incorporated, December 2000
[48]TMS320 DSP Algorithm Standard Rules and Guidelines. Texas Instruments Incorporated, 2001