基于交叉点提取的工程图表图像识别系统研究
|
摘要
扫描文档图像识别是CAD／CAM和模式识别领域中的一个重要研究方向，包括扫描工程图纸识别、扫描表格票据图像识别和工程图表图像识别等多个方面，本文主要研究工程图表图像的识别问题。
第一章对曲线图表图像计算机智能化识别的必要性进行了分析。以指导图表图像识别研究为出发点，对国内外有关扫描工程图纸和表格票据图像的矢量化识别方法进行了综述，并在此基础上建立了论文总体框架。
第二章通过分析图表识别系统的基本要求，探讨了此系统的开发环境的选择等问题。并根据需求分析和模块划分原则将系统划分为用户查询模块和图表识别模块，然后详细讨论了这两个模块的基本功能和相互关系。
第三章研究了图表图像的预处理技术，介绍了一种基于数学形态学的噪声过滤方法，并设计了四类对图表图像进行平滑的数学形态学模板算子。基于Hough变换，研究了图表图像的倾斜校正方法和图文分离方法。
第四章针对矢量化识别中出现的交叉点畸变现象，结合图表图像的结构特征，提出了基于交叉点提取的坐标线提取算法和基于斜率跟踪的曲线识别算法，分别提取图表图像中的坐标线和曲线图元，实现了图表图像矢量化识别。
第五章在分析一般印刷体数字识别技术的基础上，提出了一种基于数字质心的模板匹配数字识别算法，不经过细化，直接实现图表图像中坐标刻度的提取和定位。
第六章对图表图像识别系统的前后处理模块进行了研究。结合数字图像处理、图元编辑和曲线拟合等相关技术，实现了图表图像的图元编辑、坐标线和曲线拟合和重绘；并结合数据库技术，对用户查询模块进行了具体实现。
The recognition of scanned image of documents drawing is an important study object in fields of CAD/CAM and pattern recognition, such as engineering drawing and form drawing recognition. However, few work had been done in the past about scanned image of engineering diagram .The first chapter of this paper mainly reviews the existing research, describes the advantage and weak points of each method, and at the same time builds our own general theory framework in order to guide the research of engineering diagram recognition. According to the structure and characteristic of diagram, the pre-processing technology of diagram image, the vectorization of coordinate lines and curves in the diagram, the recognition of coordinate scale number, and user query function are discussed in detail in the remaining parts of the paper.
The second chapter gives the system design of diagram image recognition, selects a suitable developing environment by analyzing the function and capability requirement of diagram image recognition system. The whole system is divided into diagram recognition module and user query module on the basis of module division principle and fact-requirement.
The third chapter describes the preprocessing technology of diagram image recognition. A noise filtering method based on mathematical morphology is introduced. Based on HOUGH transform, the methods of diagram adjust and image-character separation is also proposed in the chapter.
The forth chapter discusses the kernel algorithm of diagram image recognition. The analysis of the recognition algorithm of engineering drawings and form drawings has showed that the distortion near cross point area is always a hard problem in drawing vectorization. So different methods are given to deal with different objects such as coordinate lines and curves in diagram iamge. To get information of coordinate lines, a vectorization method based on cross point extraction is presented creatively, and for the curve objects in the image, another adaptive tracking processing method based on slope is presented.
The fifth chapter deals with the printed digit process in diagram image. A pattern match method based on digit mass center is introduced to achieve the recognition of scale digits in diagram image.
The sixth chapter introduces the design of pre-modules and pro-modules in diagram image recognition system. Based on technology of image process, computer
graphic, and database, the design of sub-modules is discussed in detail.
第一章对曲线图表图像计算机智能化识别的必要性进行了分析。以指导图表图像识别研究为出发点，对国内外有关扫描工程图纸和表格票据图像的矢量化识别方法进行了综述，并在此基础上建立了论文总体框架。
第二章通过分析图表识别系统的基本要求，探讨了此系统的开发环境的选择等问题。并根据需求分析和模块划分原则将系统划分为用户查询模块和图表识别模块，然后详细讨论了这两个模块的基本功能和相互关系。
第三章研究了图表图像的预处理技术，介绍了一种基于数学形态学的噪声过滤方法，并设计了四类对图表图像进行平滑的数学形态学模板算子。基于Hough变换，研究了图表图像的倾斜校正方法和图文分离方法。
第四章针对矢量化识别中出现的交叉点畸变现象，结合图表图像的结构特征，提出了基于交叉点提取的坐标线提取算法和基于斜率跟踪的曲线识别算法，分别提取图表图像中的坐标线和曲线图元，实现了图表图像矢量化识别。
第五章在分析一般印刷体数字识别技术的基础上，提出了一种基于数字质心的模板匹配数字识别算法，不经过细化，直接实现图表图像中坐标刻度的提取和定位。
第六章对图表图像识别系统的前后处理模块进行了研究。结合数字图像处理、图元编辑和曲线拟合等相关技术，实现了图表图像的图元编辑、坐标线和曲线拟合和重绘；并结合数据库技术，对用户查询模块进行了具体实现。
The recognition of scanned image of documents drawing is an important study object in fields of CAD/CAM and pattern recognition, such as engineering drawing and form drawing recognition. However, few work had been done in the past about scanned image of engineering diagram .The first chapter of this paper mainly reviews the existing research, describes the advantage and weak points of each method, and at the same time builds our own general theory framework in order to guide the research of engineering diagram recognition. According to the structure and characteristic of diagram, the pre-processing technology of diagram image, the vectorization of coordinate lines and curves in the diagram, the recognition of coordinate scale number, and user query function are discussed in detail in the remaining parts of the paper.
The second chapter gives the system design of diagram image recognition, selects a suitable developing environment by analyzing the function and capability requirement of diagram image recognition system. The whole system is divided into diagram recognition module and user query module on the basis of module division principle and fact-requirement.
The third chapter describes the preprocessing technology of diagram image recognition. A noise filtering method based on mathematical morphology is introduced. Based on HOUGH transform, the methods of diagram adjust and image-character separation is also proposed in the chapter.
The forth chapter discusses the kernel algorithm of diagram image recognition. The analysis of the recognition algorithm of engineering drawings and form drawings has showed that the distortion near cross point area is always a hard problem in drawing vectorization. So different methods are given to deal with different objects such as coordinate lines and curves in diagram iamge. To get information of coordinate lines, a vectorization method based on cross point extraction is presented creatively, and for the curve objects in the image, another adaptive tracking processing method based on slope is presented.
The fifth chapter deals with the printed digit process in diagram image. A pattern match method based on digit mass center is introduced to achieve the recognition of scale digits in diagram image.
The sixth chapter introduces the design of pre-modules and pro-modules in diagram image recognition system. Based on technology of image process, computer
graphic, and database, the design of sub-modules is discussed in detail.
引文
1．戚昌滋，杨红旗，雷闻宇 机械现代设计方法学；北京：中国建筑工业出版社，1987：374～389
2. Lin X, et al. Efficient Diagram understanding with characteristic pattern detection, CVGIP, 1985,30:84～106
3. Vaxiviere P Clesstin, CAD conversion of mechanical drawings, Computer, 1992,24(7):46～54
4．易成贵，等．工程图自动识别中自适应块算法的研究．Chinagraph'96论文集，1996，529～532
5. Han C C, Fan K C.Skeleton generation of engineering drawing via contour matching. Pattern Recognition, 1994,27(2):261～275
6. Nieuwenhuizen P R V, Kiewiet O and Bronsvoort W F. An Integrated Line Tracking and Vectorization Algorithm, EUROGRAPHICS' 94,vol. 13 ,No.3,1994:c-349～c-359
7．朱建新．扫描工程图的轮廓矢量提取．工程图学学报，1997(2-3)：46～51
8. Ballard D H ,Generalization the Hough transform to detect arbitrary shapes. Pattern Recognition, 1981,13 (2): 111～122
9. Risse T. Hough transform for line recognition :Complexity of evidence accumulation and cluster detection,CVGIP,1989,46:327～345
10. Lo Rong-Chin, Tsai Wen-Hsiang. Gray-scale hough transform for thick line detection in gray-scale images. Pattern Recognition, 1995,28(5):647～661
11．李宾，谭建荣，彭群生．一个基于扫描串的统一整体矢量化算法，软件学报，1998，9(6)：426～431
12．李伟清，等．基于图段结构的整体识别方法的研究．计算机学报，1998，21(8)：753～758
13. Chen Yung-Sheng. Segmentation and association among lines and junctions for a line image. Pattern Recognition, 1994,27(9): 1135～1157
14. Tan Jianrong,Peng Qungsheng. A global approach for line recognition based on the scanned image of engineering drawings, CAD/GRAPHICS' 93,1993,2:815～821
15．李宾，谭建荣，彭群生．基于逐步求精图象数据模型的矢量化方法研究．浙江，计算机学报，1996，19(增刊)：224～231
16．武仲科，等．一种线段和圆弧的逼近方法及其在工程图纸矢量化中的应用．计算机辅助设计与图形学学报，1998，10(4)：328～332
17. Nagasamy Vijia. Engineer drawing processing and vectorization system. CVGIP, 1990,49:379～397
18. Louisa Lam, Seong-Whan Lee, Ching Y Suen. Thinning methodologies-a comprehensive survey. IEEE Transactions on Pattern Analysis Machine Intelligence, 1992,14 (9):869～885
19. Lam Louisa, Suen Ching Y. An evaluation of parallel thinning algorithm for character recognition .IEEE Transaction On PAMI, 1995,17(9):914～919
20．霍宇翔，丁宇 等．细化畸变节点形态分析及修正策略研究；北京；计算机辅助设计与图形学学报，1997，9(6)：500～505
21．霍宇翔，丁宇 等．细化后畸变节点修正修正算法的研究；清华大学学报，1997，37(8)：76～79
22. Shaoming Zhang and Seichi Nishihara. Eliminating Shape Distortions in Thinning of Engineering Drawing ,Proceedings of Second International Conference on Computer-Aided Design &Computer Graphics,Hangzhou,China, 1991,Sept,472～477
23．王咸．基于视窗理解的工程图矢量化与解析的研究[博士学位论文]，清华大学，北京，1993
24．谭柏珠，叶邦彦等．基于知识的窗口矢量化技术；北京；计算机辅助设计与图形学学报，2000，12(2)：105～109
25．章海涛，李志峰 一种基于直线提取和补全的通用表格分析方法．北京 中文信息学报 14(2)：49～54
26. Liu J, Ding X, Wu Y. Description and recognition of form and automated form data entry. In: Proceedings of the 3rd International Conference on Document Analysis and Recognition. Washington D.C.: IEEE Computer Press, 1995. 579～582
27. Fan K C, Lu J M, Wang J Y. A feature point approach to the segmentation of form documents. In: Proceedings of the 3rd International Conference on Document Analysis and Recognition. Washington D.C.: IEEE Computer Press, 1995. 623～626
28. Wang D, Srihari S N. Analysis of form images. In: Proceedings of the ist International Conference on Document Analysis and Recognition. AFCET-IRISA/INRIA, Washington D.C.: IEEE Computer Press, 1991. 181～191
29. Belaid Y, Belaid A, Turolla E. Item searching in forms: application to french tax form. In: Proceedings of the 3rd International
Conference on Document Analysis and Recognition. Washington D.C.: IEEE Computer Press, 1995. 744～747
30. Yu B, Jain A K, Generic A. System for form dropout. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1996,18(11):1127～1134
31. Hori O, Doermann D S. Robust table-form structure analysis based on box-driven reasoning. In: Proceedings of the 3rd International Conference on Document Analysis and Recognition. Washington D.C.: IEEE Computer Press, 1995. 218～221
32. Shinjo H et al. A method for connecting disappeared junction patterns on frame lines in form documents. In: Proceedings of the 4th International Conference on Document Analysis and Recognition. Washington D.C.: IEEE Computer Press, 1997. 667～670
33．李星原，高文．一种鲁棒性的结构未知表格分析方法．黑龙江，软件学报，1999，10(11)
34．徐彦欣， 彭文一．数学形态学在工程图表计算机处理中的应用．成都，西南交通大学学报．1997，32(3)：256～260
35．崔云峰，基于WEB的图表识别系统的研究[硕士学位论文]，浙江大学，2003
36．叶邦彦，谭柏珠 等．基于交叉点提取的工程图矢量化方法研究；广东；华南理工大学学报(自然科学版)，2000，28(8)：77～81
37．谭浩强 编著．C程序设计，北京：清华大学出版社，1991，7
38．王春森 等著．系统设计师(高级程序员)教程．北京 清华大学出版社2001，5
39．David J. Kruglinski, Scot Wingo, George Shepherd著．希望图书创作室Visual C++6.0技术内幕(中文第四版)北京：北京希望电子出版社，1999，5
40．Microsoft Visual Studio Co, USA MSDN Library Visual Studio 6.0版
41．张宏森，朱征宇．四层B／S结构及解决方案，成都，计算机应用研究 2002，19(9)：21～22
42．伍先俊，李志明．多层次Web应用系统的关键技术探讨，武汉：计算机工程与设计2002(6)：6～9
43．Kenneth R．Castleman著，朱志刚等译，数字图象处理，电子工业出版社，2002，2
44．韩思奇，王蕾．图像分割的阈值法综述，系统工程与电子技术，Vol．24，No．6，2002：91～95
45．张宏林编著，数字图像模式识别技术及工程实践，人民邮电出版社，2003，
2
46. Serra Jean, Image analysis and mathematical morphology, Vol.2,Academic Press, 1988
47．计算机图像与图形处理
48．张昆 顾国庆．二值图象的一种倾斜调整算法．计算机与现代化 1999，2
49．瞿洋 杨利平．Hough变换OCR图象倾斜矫正方法．西安中国图象图形学报2001，6(2)：178～181
50．何斌 马天予 等．Visual C++数字图象处理．北京人民邮电出版社．2001，4
51．《间歇运动机构设计》，上海科技出版社出版，第173页图6-4
52．叶邦彦，谭柏珠 等．基于交叉点提取的工程图矢量化方法研究；广东；华南理大学学报(自然科学版)，2000，28(8)：77～81
53. HAN C C,FAN K C,Skeleton genernation of engineering drawing via contour matching. Pattern recognition, 1994,27(2):261～275
54. KOPLOWITZ J,PLANTE S. Corner detection for chain coded curves. Pattern Recognition, 1995,28(6):843～852
55．严国莉等，印刷体数字快速识别算法在身份证编号数字识别中的应用，计算机工程，2003，29(1)：178～179
56．柳回春等，手写体数字识别技术的研究，计算机工程，2003，29(4)：24～25，61
57．许捍卫等，一种简单的数字识别方法研究，地矿测绘，2000，第二期：31，32
58．周琨等，基于模板匹配的选票图像识别研究，清华大学学报，2002，42(9)：1253～1256
59．章毓晋．图象处理和分析．北京清华大学出版社 1999，3
60．徐士良 著．C常用算法程序集 北京 清华大学出版社 1996，11．217～220
61．bbs水木清华站，电脑技术精华区 2000，telnet bbs. tsinghua. edu. cn
62. Java jdbc driver for Microsoft SQL Server 2000, http://www.inetsoftware.de
63．百联美达美公司，中国软件 csdn.net,http://www.csdn.net
64．bbs缥缈水云间站，电脑技术精华区2000，telnet 10. 13. 21. 88
2. Lin X, et al. Efficient Diagram understanding with characteristic pattern detection, CVGIP, 1985,30:84～106
3. Vaxiviere P Clesstin, CAD conversion of mechanical drawings, Computer, 1992,24(7):46～54
4．易成贵，等．工程图自动识别中自适应块算法的研究．Chinagraph'96论文集，1996，529～532
5. Han C C, Fan K C.Skeleton generation of engineering drawing via contour matching. Pattern Recognition, 1994,27(2):261～275
6. Nieuwenhuizen P R V, Kiewiet O and Bronsvoort W F. An Integrated Line Tracking and Vectorization Algorithm, EUROGRAPHICS' 94,vol. 13 ,No.3,1994:c-349～c-359
7．朱建新．扫描工程图的轮廓矢量提取．工程图学学报，1997(2-3)：46～51
8. Ballard D H ,Generalization the Hough transform to detect arbitrary shapes. Pattern Recognition, 1981,13 (2): 111～122
9. Risse T. Hough transform for line recognition :Complexity of evidence accumulation and cluster detection,CVGIP,1989,46:327～345
10. Lo Rong-Chin, Tsai Wen-Hsiang. Gray-scale hough transform for thick line detection in gray-scale images. Pattern Recognition, 1995,28(5):647～661
11．李宾，谭建荣，彭群生．一个基于扫描串的统一整体矢量化算法，软件学报，1998，9(6)：426～431
12．李伟清，等．基于图段结构的整体识别方法的研究．计算机学报，1998，21(8)：753～758
13. Chen Yung-Sheng. Segmentation and association among lines and junctions for a line image. Pattern Recognition, 1994,27(9): 1135～1157
14. Tan Jianrong,Peng Qungsheng. A global approach for line recognition based on the scanned image of engineering drawings, CAD/GRAPHICS' 93,1993,2:815～821
15．李宾，谭建荣，彭群生．基于逐步求精图象数据模型的矢量化方法研究．浙江，计算机学报，1996，19(增刊)：224～231
16．武仲科，等．一种线段和圆弧的逼近方法及其在工程图纸矢量化中的应用．计算机辅助设计与图形学学报，1998，10(4)：328～332
17. Nagasamy Vijia. Engineer drawing processing and vectorization system. CVGIP, 1990,49:379～397
18. Louisa Lam, Seong-Whan Lee, Ching Y Suen. Thinning methodologies-a comprehensive survey. IEEE Transactions on Pattern Analysis Machine Intelligence, 1992,14 (9):869～885
19. Lam Louisa, Suen Ching Y. An evaluation of parallel thinning algorithm for character recognition .IEEE Transaction On PAMI, 1995,17(9):914～919
20．霍宇翔，丁宇 等．细化畸变节点形态分析及修正策略研究；北京；计算机辅助设计与图形学学报，1997，9(6)：500～505
21．霍宇翔，丁宇 等．细化后畸变节点修正修正算法的研究；清华大学学报，1997，37(8)：76～79
22. Shaoming Zhang and Seichi Nishihara. Eliminating Shape Distortions in Thinning of Engineering Drawing ,Proceedings of Second International Conference on Computer-Aided Design &Computer Graphics,Hangzhou,China, 1991,Sept,472～477
23．王咸．基于视窗理解的工程图矢量化与解析的研究[博士学位论文]，清华大学，北京，1993
24．谭柏珠，叶邦彦等．基于知识的窗口矢量化技术；北京；计算机辅助设计与图形学学报，2000，12(2)：105～109
25．章海涛，李志峰 一种基于直线提取和补全的通用表格分析方法．北京 中文信息学报 14(2)：49～54
26. Liu J, Ding X, Wu Y. Description and recognition of form and automated form data entry. In: Proceedings of the 3rd International Conference on Document Analysis and Recognition. Washington D.C.: IEEE Computer Press, 1995. 579～582
27. Fan K C, Lu J M, Wang J Y. A feature point approach to the segmentation of form documents. In: Proceedings of the 3rd International Conference on Document Analysis and Recognition. Washington D.C.: IEEE Computer Press, 1995. 623～626
28. Wang D, Srihari S N. Analysis of form images. In: Proceedings of the ist International Conference on Document Analysis and Recognition. AFCET-IRISA/INRIA, Washington D.C.: IEEE Computer Press, 1991. 181～191
29. Belaid Y, Belaid A, Turolla E. Item searching in forms: application to french tax form. In: Proceedings of the 3rd International
Conference on Document Analysis and Recognition. Washington D.C.: IEEE Computer Press, 1995. 744～747
30. Yu B, Jain A K, Generic A. System for form dropout. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1996,18(11):1127～1134
31. Hori O, Doermann D S. Robust table-form structure analysis based on box-driven reasoning. In: Proceedings of the 3rd International Conference on Document Analysis and Recognition. Washington D.C.: IEEE Computer Press, 1995. 218～221
32. Shinjo H et al. A method for connecting disappeared junction patterns on frame lines in form documents. In: Proceedings of the 4th International Conference on Document Analysis and Recognition. Washington D.C.: IEEE Computer Press, 1997. 667～670
33．李星原，高文．一种鲁棒性的结构未知表格分析方法．黑龙江，软件学报，1999，10(11)
34．徐彦欣， 彭文一．数学形态学在工程图表计算机处理中的应用．成都，西南交通大学学报．1997，32(3)：256～260
35．崔云峰，基于WEB的图表识别系统的研究[硕士学位论文]，浙江大学，2003
36．叶邦彦，谭柏珠 等．基于交叉点提取的工程图矢量化方法研究；广东；华南理工大学学报(自然科学版)，2000，28(8)：77～81
37．谭浩强 编著．C程序设计，北京：清华大学出版社，1991，7
38．王春森 等著．系统设计师(高级程序员)教程．北京 清华大学出版社2001，5
39．David J. Kruglinski, Scot Wingo, George Shepherd著．希望图书创作室Visual C++6.0技术内幕(中文第四版)北京：北京希望电子出版社，1999，5
40．Microsoft Visual Studio Co, USA MSDN Library Visual Studio 6.0版
41．张宏森，朱征宇．四层B／S结构及解决方案，成都，计算机应用研究 2002，19(9)：21～22
42．伍先俊，李志明．多层次Web应用系统的关键技术探讨，武汉：计算机工程与设计2002(6)：6～9
43．Kenneth R．Castleman著，朱志刚等译，数字图象处理，电子工业出版社，2002，2
44．韩思奇，王蕾．图像分割的阈值法综述，系统工程与电子技术，Vol．24，No．6，2002：91～95
45．张宏林编著，数字图像模式识别技术及工程实践，人民邮电出版社，2003，
2
46. Serra Jean, Image analysis and mathematical morphology, Vol.2,Academic Press, 1988
47．计算机图像与图形处理
48．张昆 顾国庆．二值图象的一种倾斜调整算法．计算机与现代化 1999，2
49．瞿洋 杨利平．Hough变换OCR图象倾斜矫正方法．西安中国图象图形学报2001，6(2)：178～181
50．何斌 马天予 等．Visual C++数字图象处理．北京人民邮电出版社．2001，4
51．《间歇运动机构设计》，上海科技出版社出版，第173页图6-4
52．叶邦彦，谭柏珠 等．基于交叉点提取的工程图矢量化方法研究；广东；华南理大学学报(自然科学版)，2000，28(8)：77～81
53. HAN C C,FAN K C,Skeleton genernation of engineering drawing via contour matching. Pattern recognition, 1994,27(2):261～275
54. KOPLOWITZ J,PLANTE S. Corner detection for chain coded curves. Pattern Recognition, 1995,28(6):843～852
55．严国莉等，印刷体数字快速识别算法在身份证编号数字识别中的应用，计算机工程，2003，29(1)：178～179
56．柳回春等，手写体数字识别技术的研究，计算机工程，2003，29(4)：24～25，61
57．许捍卫等，一种简单的数字识别方法研究，地矿测绘，2000，第二期：31，32
58．周琨等，基于模板匹配的选票图像识别研究，清华大学学报，2002，42(9)：1253～1256
59．章毓晋．图象处理和分析．北京清华大学出版社 1999，3
60．徐士良 著．C常用算法程序集 北京 清华大学出版社 1996，11．217～220
61．bbs水木清华站，电脑技术精华区 2000，telnet bbs. tsinghua. edu. cn
62. Java jdbc driver for Microsoft SQL Server 2000, http://www.inetsoftware.de
63．百联美达美公司，中国软件 csdn.net,http://www.csdn.net
64．bbs缥缈水云间站，电脑技术精华区2000，telnet 10. 13. 21. 88
© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号 地址：北京市海淀区学院路29号 邮编：100083 电话：办公室：(+86 10)66554848；文献借阅、咨询服务、科技查新：66554700 |