车牌定位与字符分割的研究与实现
|
摘要
进入新世纪，随着我国公路交通事业的快速发展，对智能交通系统的需求又进入新的阶段。车辆本身的信息，如车牌号，车主，养路费的征收情况等是车辆自动化管理的操作对象，而车牌识别是车辆信息的获取的最主要的手段。由此可见，车牌识别系统在智能交通领域的重要作用。
一个完整的车牌识别系统由车牌定位、字符分割和字符识别等单元组成，涉及面广。其中车牌定位和字符分割在整个系统中起着关键性作用，它们的处理结果直接影响整个系统的精确度，鉴于此本文研究的主要内容是车牌定位与字符分割的方法。
本文中研究了两种车牌的定位方法：第一种是基于连通域分析的车牌定位方法。在选定的阈值范围内二值化图像，穷举其连通域轮廓，借助链码计算连通域的形状参数，筛选轮廓形状并以过滤后的轮廓中心线确定车牌位置；第二种方法基于图像的垂直边缘特征，使用垂直方向的Prewitt梯度算子提取图像的垂直边缘，逐点以边缘点密度为衡量标准来优化边缘特征图像，最后用给定矩形面积内的击中率来确定车牌的位置。
本文在字符分割方面结合水平投影法和固定间距法，研究一种复合的分割方法，实验证明有较好地分割效果。
文中还讨论了基于Hough变换的车牌矫正方法，运用Hough变换检测车牌在垂直和水平方向上的倾斜角，较好地矫正了旋转和变形车牌。
As coming along with the development of the highway traffic enterprise in the new century, the requirement of the Intelligent Traffic System (ITS) gets into a new level. The information about vehicle's manufacturer, the owner, and the road maintenance cost and so on are the objects of the auto management of vehicle information, and one of the useful method to get these information is by Vehicle License Recognition System. Thus, Vehicle License Recognition system play a vital role in ITS.
A Vehicle License Recognition system usually comprises three parts, vehicle license locating, character segmentation and character recognition. Vehicle license locating and character segmentation play the crucial role, which influence the precision of the whole system directly. Thus, this thesis mainly studies the vehicle license location and the character division.
In this thesis, two vehicle licenses locating methods are studied. The first one is based on the analysis of region connectivity, which transforms image into binary images in continuous thresholds, getting all contours by tracing algorithm, and all the licenses plate liked regions are selected to compute the center position of the license plate. The second one is based on edge detection, by through extracting vertical edges to generate the density of edge points in a rectangle region, and the region with the most density can be considered as the position of license plate.
In this thesis, area projection and fixed space gap are adapted to character segmentation, and experiments shows a good result.
In this thesis, one method that uses the Hough Transform to correct the license plate is discussed also. By using the Hough Transform to compute incline angles of the plate in horizontal and vertical directions. The distort license plate can be corrected well.
一个完整的车牌识别系统由车牌定位、字符分割和字符识别等单元组成，涉及面广。其中车牌定位和字符分割在整个系统中起着关键性作用，它们的处理结果直接影响整个系统的精确度，鉴于此本文研究的主要内容是车牌定位与字符分割的方法。
本文中研究了两种车牌的定位方法：第一种是基于连通域分析的车牌定位方法。在选定的阈值范围内二值化图像，穷举其连通域轮廓，借助链码计算连通域的形状参数，筛选轮廓形状并以过滤后的轮廓中心线确定车牌位置；第二种方法基于图像的垂直边缘特征，使用垂直方向的Prewitt梯度算子提取图像的垂直边缘，逐点以边缘点密度为衡量标准来优化边缘特征图像，最后用给定矩形面积内的击中率来确定车牌的位置。
本文在字符分割方面结合水平投影法和固定间距法，研究一种复合的分割方法，实验证明有较好地分割效果。
文中还讨论了基于Hough变换的车牌矫正方法，运用Hough变换检测车牌在垂直和水平方向上的倾斜角，较好地矫正了旋转和变形车牌。
As coming along with the development of the highway traffic enterprise in the new century, the requirement of the Intelligent Traffic System (ITS) gets into a new level. The information about vehicle's manufacturer, the owner, and the road maintenance cost and so on are the objects of the auto management of vehicle information, and one of the useful method to get these information is by Vehicle License Recognition System. Thus, Vehicle License Recognition system play a vital role in ITS.
A Vehicle License Recognition system usually comprises three parts, vehicle license locating, character segmentation and character recognition. Vehicle license locating and character segmentation play the crucial role, which influence the precision of the whole system directly. Thus, this thesis mainly studies the vehicle license location and the character division.
In this thesis, two vehicle licenses locating methods are studied. The first one is based on the analysis of region connectivity, which transforms image into binary images in continuous thresholds, getting all contours by tracing algorithm, and all the licenses plate liked regions are selected to compute the center position of the license plate. The second one is based on edge detection, by through extracting vertical edges to generate the density of edge points in a rectangle region, and the region with the most density can be considered as the position of license plate.
In this thesis, area projection and fixed space gap are adapted to character segmentation, and experiments shows a good result.
In this thesis, one method that uses the Hough Transform to correct the license plate is discussed also. By using the Hough Transform to compute incline angles of the plate in horizontal and vertical directions. The distort license plate can be corrected well.
引文
1．陆化普，李瑞敏，朱茵．智能交通系统概述．第1版．北京：中国铁道出版社，2004
2．陆化普，智能交通系统，第1版．北京：人民交通出版社，2003
3．Milan Sonka，Vaclav Hlavac，Roger Boyle．艾海舟，武勃等译．图像处理、分析与机器视觉．第2版．北京：人民邮电出版社，2003．9
4．徐建华．图像处理与分析．第1版．北京：科学出版社，1992．1
5．任明武．数字图像处理．第1版．南京：南京理工大学出版社，2003
6．杨静宇，曹雨龙．计算机图像处理及常用算法手册．第1版．南京：南京大学出版社．1997
7．潘晓露，李一民．中国交通信息产业．2003
8．荆仁杰．计算机图像处理．浙江：浙江大学出版社．1992
9．夏良正．数字图像处理．南京：东南大学出版社．1999
10．王润生．图像理解．长沙：国防科技大学出版社．1995
11．吴健康．数字图像分析．北京：人民邮电出版社．1999
12．章毓晋．图像处理和分析．北京：清华大学出版社．2003
13．夏德深，傅德胜．现代图像处理技术及应用．南京：东南大学出版社．1997
14．陆系群，陈纯．图像处理原理、技术与算法．浙江：浙江大学出版社．2001
15．张学工等．模式识别．北京：清华大学出版社．2000
16．唐常青。数学形态学方法及其应用．北京：科学出版社．1990
17．崔屹．图像处理与分析．北京：科学出版社．2000
18．李介谷．计算机视觉的理论和实践．上海：上海交通大学出版社．1999
19．阎建国，高亮，卢京潮．图像处理技术在车牌识别中的应用．计算机应用，2000，22(1)：17-18
20. Huang T S, Tang G T.A fast two-dimensional median filtering algorithm. IEEE Trans Acoustics, Speech, and Signal Processing, 1979, 27(1): 1318
21. Gordon R, Rangayyan R M. Feature enhancement of film mammograms using fixed and adaptive neighborhoods. Applied Optics, 1984, 3(4): 560-574
22. Mingwu Ren, Jingyu Yang, Han sun, Tracing Boundary Contours in a Binary Image. Image and Vision Computing, 20(2), pp. 125-131, 2002
23. Freeman H. On the encoding of arbitrary geometric configuration. IRE Trans, 1961, 10(2): 260～268
24. Y T Liow. A contour tracing algorithm that preserves common boundaries between regions. CVGIP-Image Understanding, 53(3):313-321, 1991.
25．赵巨波，孙华燕，杜巍．一种图像边缘特征提取算法．光学精密工程，2000，8(4)：325-327
26．丁秋芳．车牌自动定位切分技术．江苏交通．2003，1
27．张引，潘云鹤．面向车辆牌照字符识别的预处理算法．计算机应用研究．1997，7．85-87
28．廖金周，宣国荣．车辆牌照的自动分割．微型电脑应用．1999．7：32-34
29．李炜，黄心汉，王敏，陈曦，魏武．车牌自动识别系统中车牌分割的研究．信号处理，2000．16：41～44
30．叶晨洲，廖金周等．车辆牌照字符识别系统．计算机系统应用．1999．05：10-13
31．牛欣，沈兰荪．汽车牌照识别技术的研究．测控技术，1999，18(12)：14～17
32．魏武，黄心汉．MBEDE Jean-boscol等．一种基于垂直字符边界特征的车牌定位方法．中国公路学报，2000，13(4)：88～90
33. Mei Yu, Yong Deak Kim. An approach to Korean license plate recognition based on vertical edge matching. In:Proceedings of IEEE International Conference on Systems, Man and Cybernetics, 2000,4:2975～2980
34．张西宁，郑南宁．汽车牌照自动识别中的目标与背景的快速分割．信息与控制．1988，(2)：27-31
35．王少杰，朱治刚，石定机，俞志和．货运列车车型车号自动分割和识别算法．模式识别与人工智能，1998．(3)：328-334．
36．沈定刚等。汽车牌照自动定位算法．无线电通信技术，1995，21(5)：35-38
37. Thanongsak Sirithinaphong, Kosin Chamnongthai. The Recognition of Car License Hate for Automatic Parking System. Proceedings of the Fifth International Sympo-sium on Signal Processing and Its Applications, 1999 (1):455-457
38. M H ter Brugge, J H Stevens, J A G Nijhuis, L Spaanenburg. License Plate Recognition Using DTC-NNs. Proceedings of the IEEE International Workshopon Cellular Neural Networks and their Applications, 1998, 212-217
2．陆化普，智能交通系统，第1版．北京：人民交通出版社，2003
3．Milan Sonka，Vaclav Hlavac，Roger Boyle．艾海舟，武勃等译．图像处理、分析与机器视觉．第2版．北京：人民邮电出版社，2003．9
4．徐建华．图像处理与分析．第1版．北京：科学出版社，1992．1
5．任明武．数字图像处理．第1版．南京：南京理工大学出版社，2003
6．杨静宇，曹雨龙．计算机图像处理及常用算法手册．第1版．南京：南京大学出版社．1997
7．潘晓露，李一民．中国交通信息产业．2003
8．荆仁杰．计算机图像处理．浙江：浙江大学出版社．1992
9．夏良正．数字图像处理．南京：东南大学出版社．1999
10．王润生．图像理解．长沙：国防科技大学出版社．1995
11．吴健康．数字图像分析．北京：人民邮电出版社．1999
12．章毓晋．图像处理和分析．北京：清华大学出版社．2003
13．夏德深，傅德胜．现代图像处理技术及应用．南京：东南大学出版社．1997
14．陆系群，陈纯．图像处理原理、技术与算法．浙江：浙江大学出版社．2001
15．张学工等．模式识别．北京：清华大学出版社．2000
16．唐常青。数学形态学方法及其应用．北京：科学出版社．1990
17．崔屹．图像处理与分析．北京：科学出版社．2000
18．李介谷．计算机视觉的理论和实践．上海：上海交通大学出版社．1999
19．阎建国，高亮，卢京潮．图像处理技术在车牌识别中的应用．计算机应用，2000，22(1)：17-18
20. Huang T S, Tang G T.A fast two-dimensional median filtering algorithm. IEEE Trans Acoustics, Speech, and Signal Processing, 1979, 27(1): 1318
21. Gordon R, Rangayyan R M. Feature enhancement of film mammograms using fixed and adaptive neighborhoods. Applied Optics, 1984, 3(4): 560-574
22. Mingwu Ren, Jingyu Yang, Han sun, Tracing Boundary Contours in a Binary Image. Image and Vision Computing, 20(2), pp. 125-131, 2002
23. Freeman H. On the encoding of arbitrary geometric configuration. IRE Trans, 1961, 10(2): 260～268
24. Y T Liow. A contour tracing algorithm that preserves common boundaries between regions. CVGIP-Image Understanding, 53(3):313-321, 1991.
25．赵巨波，孙华燕，杜巍．一种图像边缘特征提取算法．光学精密工程，2000，8(4)：325-327
26．丁秋芳．车牌自动定位切分技术．江苏交通．2003，1
27．张引，潘云鹤．面向车辆牌照字符识别的预处理算法．计算机应用研究．1997，7．85-87
28．廖金周，宣国荣．车辆牌照的自动分割．微型电脑应用．1999．7：32-34
29．李炜，黄心汉，王敏，陈曦，魏武．车牌自动识别系统中车牌分割的研究．信号处理，2000．16：41～44
30．叶晨洲，廖金周等．车辆牌照字符识别系统．计算机系统应用．1999．05：10-13
31．牛欣，沈兰荪．汽车牌照识别技术的研究．测控技术，1999，18(12)：14～17
32．魏武，黄心汉．MBEDE Jean-boscol等．一种基于垂直字符边界特征的车牌定位方法．中国公路学报，2000，13(4)：88～90
33. Mei Yu, Yong Deak Kim. An approach to Korean license plate recognition based on vertical edge matching. In:Proceedings of IEEE International Conference on Systems, Man and Cybernetics, 2000,4:2975～2980
34．张西宁，郑南宁．汽车牌照自动识别中的目标与背景的快速分割．信息与控制．1988，(2)：27-31
35．王少杰，朱治刚，石定机，俞志和．货运列车车型车号自动分割和识别算法．模式识别与人工智能，1998．(3)：328-334．
36．沈定刚等。汽车牌照自动定位算法．无线电通信技术，1995，21(5)：35-38
37. Thanongsak Sirithinaphong, Kosin Chamnongthai. The Recognition of Car License Hate for Automatic Parking System. Proceedings of the Fifth International Sympo-sium on Signal Processing and Its Applications, 1999 (1):455-457
38. M H ter Brugge, J H Stevens, J A G Nijhuis, L Spaanenburg. License Plate Recognition Using DTC-NNs. Proceedings of the IEEE International Workshopon Cellular Neural Networks and their Applications, 1998, 212-217