车牌定位技术的研究
|
摘要
随着汽车数量的日益增多,城市的交通状况、停车场的管理等不可避免受到人们重视。车辆牌照作为汽车的重要特征之一,它的识别不可避免的成为智能交通系统的核心,起着举足轻重的作用。
本文对车牌识别的几个关键技术进行了研究,并对各个关键技术进行了理论分析和仿真实验。主要研究工作如下:
(1)车辆定位。车辆定位主要有两个方面的原因:一个是可以缩小车牌识别的范围,另一个可以实现复杂场景监控。本文做了基于序列图像的运动车辆定位和基于深度信息的车辆定位。基于深度信息的车辆定位利用了深度图的特性。由于采用了多种探测手段,有效的恢复了场景的空间和颜色信息。根据场景的深度信息,实现了目标与背景的分离,从而消除了背景对前景目标的干扰。对两种方法实验结果的比较,基于深度信息的运动车辆定位提高了前景目标的检测速度和背景的更新速度,并且前景目标检测有不受阴影、光照变化的影响等特点。
(2)车牌定位。车牌定位是车牌自动识别系统的关键技术,如果不能在图像中正确找到车牌的位置,车牌号码的识别就无法进行。车牌定位主要通过车牌区域的特征来实现,本文主要研究通过形态学实现对车牌区域的定位,然后通过车牌区域的特征来对车牌候选区域进行验证。
实验结果表明,本系统提出的基于深度信息的车辆定位具有很好的准确性,车牌定位精度也比较高。
With the growing number of cars,People also pay more attention on city's traffic conditions,the management of car parks and so on.License Plate as one of the most important feature of car,its Recognition inevitable become the heart of intelligent traffic system,and play an important role in ITS.
This article mainly research some important technology of license plate recognition,illustrator every important technology by theoretical analysis and simulation experiment,author's main research work as follows:
(1) Car Positioning.There are two main reasons for car positioning:the one is that it can narrow the scope for license plate recognition,the other is that it can monitor a complex scene.This article illustrator two methods to achieve car positioning,one is based on image sequences,the other is based on depth information. the one based on depth information use the feature of range image.As a result of many detection methods,the scenes space and color information are regained effectively.According to the depth information of a scene,Ⅰachieve the separation of the foreground target and the background.Eliminate the interference on the background by foreground target.Compare the results of the two methods,the one based on depth information enhance the target detection rate and the background update rate,shadows and light have nothing impacts on foreground target detection, finally achieve the car positioning.
(2) License Plate Positioning.License plate positioning is the very important technology of the license plate intelligent recognition system,if we can't find the right position of the license plate,license number recognition is unable to carry out. License plate positioning primarily through the characteristics of the region to achieve.In this article,Ⅰmainly research how to use morphology,two-valued projection histogram analysis to achieve the license plate positioning.
本文对车牌识别的几个关键技术进行了研究,并对各个关键技术进行了理论分析和仿真实验。主要研究工作如下:
(1)车辆定位。车辆定位主要有两个方面的原因:一个是可以缩小车牌识别的范围,另一个可以实现复杂场景监控。本文做了基于序列图像的运动车辆定位和基于深度信息的车辆定位。基于深度信息的车辆定位利用了深度图的特性。由于采用了多种探测手段,有效的恢复了场景的空间和颜色信息。根据场景的深度信息,实现了目标与背景的分离,从而消除了背景对前景目标的干扰。对两种方法实验结果的比较,基于深度信息的运动车辆定位提高了前景目标的检测速度和背景的更新速度,并且前景目标检测有不受阴影、光照变化的影响等特点。
(2)车牌定位。车牌定位是车牌自动识别系统的关键技术,如果不能在图像中正确找到车牌的位置,车牌号码的识别就无法进行。车牌定位主要通过车牌区域的特征来实现,本文主要研究通过形态学实现对车牌区域的定位,然后通过车牌区域的特征来对车牌候选区域进行验证。
实验结果表明,本系统提出的基于深度信息的车辆定位具有很好的准确性,车牌定位精度也比较高。
With the growing number of cars,People also pay more attention on city's traffic conditions,the management of car parks and so on.License Plate as one of the most important feature of car,its Recognition inevitable become the heart of intelligent traffic system,and play an important role in ITS.
This article mainly research some important technology of license plate recognition,illustrator every important technology by theoretical analysis and simulation experiment,author's main research work as follows:
(1) Car Positioning.There are two main reasons for car positioning:the one is that it can narrow the scope for license plate recognition,the other is that it can monitor a complex scene.This article illustrator two methods to achieve car positioning,one is based on image sequences,the other is based on depth information. the one based on depth information use the feature of range image.As a result of many detection methods,the scenes space and color information are regained effectively.According to the depth information of a scene,Ⅰachieve the separation of the foreground target and the background.Eliminate the interference on the background by foreground target.Compare the results of the two methods,the one based on depth information enhance the target detection rate and the background update rate,shadows and light have nothing impacts on foreground target detection, finally achieve the car positioning.
(2) License Plate Positioning.License plate positioning is the very important technology of the license plate intelligent recognition system,if we can't find the right position of the license plate,license number recognition is unable to carry out. License plate positioning primarily through the characteristics of the region to achieve.In this article,Ⅰmainly research how to use morphology,two-valued projection histogram analysis to achieve the license plate positioning.
引文
[1]章毓晋.图像图形科学丛书--图像分割.北京:科学出版社,2001.
[2]何东健.数字图像处理.西安:西安电子科技大学出版社,2003.17-33.
[3]Kerry,C G,richard E W.Digital Image Processing[M].Beijing:Publishing House of Electronics Industry,2003:282-348,519-566.
[4]白洪亮、娄正良、邹明福.等.复杂背景下基于形态学的车牌识别系统.公路交通科技.2004。121(110):117-120
[5]Aghdasi,F,Ndungo,H.Automatic license plate recognition system.AFRICON,2004.7AFRICON Conference in Africa Volume 1,15-17 Sept.2004 Page(s):45 - 50 Vol.1.
[6]Yoshimod,S.Mitsukura,Y.Funkumi,M.Akarnatsu,N.Khosal,R.License plate detection system in rainy days.Computational Intelligence in Robotics and Automation,16-20 July 2003,972 - 976 vol.2
[7]Nathan,Vidhya Sudhan L,Kamakshi Priya.New Approaches for License Plate Recognition System.Proceedings of International Conference on Intelligent Sensing and Information Processing.2004:149-152.
[8]Kawaguchi H.Application system using license plate recognition technology[J].NEC Techical journal,2005,54(7):19-22.
[9]刘焕军,王耀南,段峰.机器礼堂中的图像采集技术.计算机与信息技,2003(1):18-21
[10]李波,曾致远,付祥胜,基于数学形态学和边缘特征的车牌定位算法,视频技术应用与工程,2005,(7):94-96
[11]袁宝民、于万波、魏小鹏,汽车牌照定位研究综述,大连大学学报,2002.23(2).
[12]路小波、张光华,基于二值图像的车牌精确定位方法,东南大学学报,2005年11月
[13]Chen Xiangrong,A.L.Yuille.Detecting and Reading Text in Natural Scenes.Proceedings of the IEEE Computer Society Conference,2004:366-373.
[14]Da Rocha Gesualdi,A.Manoelde Seixas,J.C Character recognition incar license plates based 2002 SRRN on principal 2002.components and neural processing,Neural Networks,
[15]Shen-Zheng Wang His-Jian Lee,Detection and recognition of license plate Character swith different appearances,Intelligent Transportation Systems,2003
[16]Ko,Mi-Ae,Kim,Young-Mo.Improvements in Real-time Feature-based License Plate Character Recognition.Key Engineering materials,2005(1):355-360.
[17]Obeid,Hasan R.Zantout etc,License Processing:Recognizing Characters in License Plates.2007 IEEE/ACS International Conference on Computer Systems and Applications,2007:712-716.
[18]代科学,李国辉,涂丹,等.监控视频运动目标检测减背景技术的研究现状和展望.中国图像图形学报[J],2006,7(11):919-927.
[19]马艳,张治辉.几种边缘检测操作数的比较.工矿自动化,2004,1(2):54-56.
[20]Li Li-yuan Gu Irene Yu-Hua,Leung Maylor K H,etal,Adaptive background subtraction based on feedback from fuzzy classification[J].Optical Engineering,2004,43(10):2381-2394.
[21]Salvador Elena,Cavallaro Andrea,Ebrahimi Touradj.Cast shadow segmentation using invariant color features[J].Computer Vision and Image Understanding,2004,95(2):238-259
[22]Bremananth,R,Nathan,Vidya,Sudhan Letal.A New Methodology for License Plate Recognition IEEE Journal of Research,2005(51):279-286.
[23]穆长江、苑玮琦,基于纹理牲的车牌定位方法,控制工程,2004年11月
[24]郭捷,郭鹏飞,基于颜色和纹理分析的车牌定位方法,中国图像图形学报,2002,(7):472-476
[25]Kahl F,Hartley R,Hilsenstein V.Novelty detection in image sequences with dynamic background[C].In:Statistical Methods in video Processing,Berlin:Springer-Verlag Berlin,2004:117-128.
[26]Takahashi,Yasuko,Tanaka,Hisako etc,License Plate Recognition using Gray Image template Matching with Noise Reduction Filters and Character Alignment.System and Computers in Japan,2007(3):49-61.
[27]左奇,史忠科.一种基于数学形态学的实时车牌图像分割方法.中国图像图形学报,2003,3(8):281-285
[28]葛庆国,吴晓娟,江东梅.改进的基于区域的运动目标分割方法.计算机工程,2004,12(30):106-108
[29]Sunghoom Kim,Daeehul Kim,Younbok Ryu,Gyeonghwan Kim.A robust license-plate extraction method under complex image conditions.Pattern Recognition,2002.Proceedings.16~(th) International conference,11 -15 Aug.2002,216 - 219 vol.3.
[30]Comelli P,Ferragina P.Optical Recognition of Motor vehicle Licence Plates[J]IEEE Transction on Vehicle Technology,2003,11(44):790-792.
[31]Stauffer Chris,Grimson WEL.Adaptive background mixture models for real-time tracking[J],Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition,1999,2:246-252.
[32]C Rittscher,J Kato.A Probabilistic Background Model for Tracking.Eccv 2000,Vol.2.
[33]Shyang-Lih Chang,Li-Shien Chen,Yun-Chung,Sei-Wan Chen.Automatic license plate
[34]Oz,Cemil;Ercal,Fikret.A Practical License Plate Recognition System for Real-time Environments.8~(th) International Workshop on Artificial Neural Networks,2005(35):881-888.
[35]Barroso,A Rafael.Number Plate Reading Using Computer Vision.IEEE Transaction on Vehicular Technology.2001,11.
[36]Parist R.Car Plate Recognition by Neural Networks and Image Processing,IEEE International Symposium on Circuits and systems,USA,2000,May 31-June 3.
[37]P.Viola,M.J.Jones.Robust Real-Time Face Detection.International Journal of Computer Vision 2004,Vol.57,No.2:137-154.
[38]Yohimori,Mitsukura,Fukumi etc.License Plate Detection System by using Threshold Function and Improved Template Matching Method.2004 Annual Meeting of the North American Fuzzy Information Processing Society:Fuzzy Sets in the Heart of the Canadian Rockies,204:357-362.
[39]孙惠,周红霞,李朝辉.图像处理中边缘检测技术的研究[J].计算机开发与应用,2002,15(10):1-3.
[40]肖梅、韩祟昭、张雷,基于多尺度轮廓结构元素的数学形态学边缘检测,西安交通大学学报,2005年6月.
[2]何东健.数字图像处理.西安:西安电子科技大学出版社,2003.17-33.
[3]Kerry,C G,richard E W.Digital Image Processing[M].Beijing:Publishing House of Electronics Industry,2003:282-348,519-566.
[4]白洪亮、娄正良、邹明福.等.复杂背景下基于形态学的车牌识别系统.公路交通科技.2004。121(110):117-120
[5]Aghdasi,F,Ndungo,H.Automatic license plate recognition system.AFRICON,2004.7AFRICON Conference in Africa Volume 1,15-17 Sept.2004 Page(s):45 - 50 Vol.1.
[6]Yoshimod,S.Mitsukura,Y.Funkumi,M.Akarnatsu,N.Khosal,R.License plate detection system in rainy days.Computational Intelligence in Robotics and Automation,16-20 July 2003,972 - 976 vol.2
[7]Nathan,Vidhya Sudhan L,Kamakshi Priya.New Approaches for License Plate Recognition System.Proceedings of International Conference on Intelligent Sensing and Information Processing.2004:149-152.
[8]Kawaguchi H.Application system using license plate recognition technology[J].NEC Techical journal,2005,54(7):19-22.
[9]刘焕军,王耀南,段峰.机器礼堂中的图像采集技术.计算机与信息技,2003(1):18-21
[10]李波,曾致远,付祥胜,基于数学形态学和边缘特征的车牌定位算法,视频技术应用与工程,2005,(7):94-96
[11]袁宝民、于万波、魏小鹏,汽车牌照定位研究综述,大连大学学报,2002.23(2).
[12]路小波、张光华,基于二值图像的车牌精确定位方法,东南大学学报,2005年11月
[13]Chen Xiangrong,A.L.Yuille.Detecting and Reading Text in Natural Scenes.Proceedings of the IEEE Computer Society Conference,2004:366-373.
[14]Da Rocha Gesualdi,A.Manoelde Seixas,J.C Character recognition incar license plates based 2002 SRRN on principal 2002.components and neural processing,Neural Networks,
[15]Shen-Zheng Wang His-Jian Lee,Detection and recognition of license plate Character swith different appearances,Intelligent Transportation Systems,2003
[16]Ko,Mi-Ae,Kim,Young-Mo.Improvements in Real-time Feature-based License Plate Character Recognition.Key Engineering materials,2005(1):355-360.
[17]Obeid,Hasan R.Zantout etc,License Processing:Recognizing Characters in License Plates.2007 IEEE/ACS International Conference on Computer Systems and Applications,2007:712-716.
[18]代科学,李国辉,涂丹,等.监控视频运动目标检测减背景技术的研究现状和展望.中国图像图形学报[J],2006,7(11):919-927.
[19]马艳,张治辉.几种边缘检测操作数的比较.工矿自动化,2004,1(2):54-56.
[20]Li Li-yuan Gu Irene Yu-Hua,Leung Maylor K H,etal,Adaptive background subtraction based on feedback from fuzzy classification[J].Optical Engineering,2004,43(10):2381-2394.
[21]Salvador Elena,Cavallaro Andrea,Ebrahimi Touradj.Cast shadow segmentation using invariant color features[J].Computer Vision and Image Understanding,2004,95(2):238-259
[22]Bremananth,R,Nathan,Vidya,Sudhan Letal.A New Methodology for License Plate Recognition IEEE Journal of Research,2005(51):279-286.
[23]穆长江、苑玮琦,基于纹理牲的车牌定位方法,控制工程,2004年11月
[24]郭捷,郭鹏飞,基于颜色和纹理分析的车牌定位方法,中国图像图形学报,2002,(7):472-476
[25]Kahl F,Hartley R,Hilsenstein V.Novelty detection in image sequences with dynamic background[C].In:Statistical Methods in video Processing,Berlin:Springer-Verlag Berlin,2004:117-128.
[26]Takahashi,Yasuko,Tanaka,Hisako etc,License Plate Recognition using Gray Image template Matching with Noise Reduction Filters and Character Alignment.System and Computers in Japan,2007(3):49-61.
[27]左奇,史忠科.一种基于数学形态学的实时车牌图像分割方法.中国图像图形学报,2003,3(8):281-285
[28]葛庆国,吴晓娟,江东梅.改进的基于区域的运动目标分割方法.计算机工程,2004,12(30):106-108
[29]Sunghoom Kim,Daeehul Kim,Younbok Ryu,Gyeonghwan Kim.A robust license-plate extraction method under complex image conditions.Pattern Recognition,2002.Proceedings.16~(th) International conference,11 -15 Aug.2002,216 - 219 vol.3.
[30]Comelli P,Ferragina P.Optical Recognition of Motor vehicle Licence Plates[J]IEEE Transction on Vehicle Technology,2003,11(44):790-792.
[31]Stauffer Chris,Grimson WEL.Adaptive background mixture models for real-time tracking[J],Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition,1999,2:246-252.
[32]C Rittscher,J Kato.A Probabilistic Background Model for Tracking.Eccv 2000,Vol.2.
[33]Shyang-Lih Chang,Li-Shien Chen,Yun-Chung,Sei-Wan Chen.Automatic license plate
[34]Oz,Cemil;Ercal,Fikret.A Practical License Plate Recognition System for Real-time Environments.8~(th) International Workshop on Artificial Neural Networks,2005(35):881-888.
[35]Barroso,A Rafael.Number Plate Reading Using Computer Vision.IEEE Transaction on Vehicular Technology.2001,11.
[36]Parist R.Car Plate Recognition by Neural Networks and Image Processing,IEEE International Symposium on Circuits and systems,USA,2000,May 31-June 3.
[37]P.Viola,M.J.Jones.Robust Real-Time Face Detection.International Journal of Computer Vision 2004,Vol.57,No.2:137-154.
[38]Yohimori,Mitsukura,Fukumi etc.License Plate Detection System by using Threshold Function and Improved Template Matching Method.2004 Annual Meeting of the North American Fuzzy Information Processing Society:Fuzzy Sets in the Heart of the Canadian Rockies,204:357-362.
[39]孙惠,周红霞,李朝辉.图像处理中边缘检测技术的研究[J].计算机开发与应用,2002,15(10):1-3.
[40]肖梅、韩祟昭、张雷,基于多尺度轮廓结构元素的数学形态学边缘检测,西安交通大学学报,2005年6月.