中文版面分析的研究
|
摘要
版面分析作为文字识别系统的预处理部分，其准确性直接影响文字的识别率。本文针对复杂的中文版面，提出了一个基于模糊连接度和识别特征的中文版面分析方法，完成了一个图像输入、倾斜校正、版面图文分割的过程。版面图文分割主要采用自底向上的办法，利用连通域搜索算法检测出文本页面上的所有连通基元，通过对连通基元的四个方向上的连接度进行模糊化处理来决定文字行、列的合并，并对在文字行合并时影响较大的标点符号采用先识别后合并的方法。为了减少时间开销，在计算和合并过程中采用局部搜索策略。实验结果表明，该方法对印刷质量比较好的中文版面具有较理想的分割效果。
The layout analysis is part of important pre-processing of character recognition. The accuracy of Layout analysis has direct effect on efficiency of character. We provide a Chinese layout analysis method base on fuzzy connectedness and recognition features for complex document layout. This is a process including the input document images ., skew correction and Texts/Graphics segmentation. The bottom-up approach used in Texts/Graphics segmentation. All the connected units in the page are detected by search algorithm of connected region. The row-column mergence of the character is defined by fuzzy connectedness of the connected units at four orientations. The combination of punctuation we adopt the method of combination behind recognition due to great effect of mergence. In order to reduce time overhead, the local searching strategy is used in the process of calculation and mergence. The result of experiment has shown that this method can analysis belter prinled-quality document with satisfactory segmentation.
The layout analysis is part of important pre-processing of character recognition. The accuracy of Layout analysis has direct effect on efficiency of character. We provide a Chinese layout analysis method base on fuzzy connectedness and recognition features for complex document layout. This is a process including the input document images ., skew correction and Texts/Graphics segmentation. The bottom-up approach used in Texts/Graphics segmentation. All the connected units in the page are detected by search algorithm of connected region. The row-column mergence of the character is defined by fuzzy connectedness of the connected units at four orientations. The combination of punctuation we adopt the method of combination behind recognition due to great effect of mergence. In order to reduce time overhead, the local searching strategy is used in the process of calculation and mergence. The result of experiment has shown that this method can analysis belter prinled-quality document with satisfactory segmentation.
引文
[1] 姜哲，夏莹．中文版面分析技术．第六届全国汉字识别学术会议论文集．1996，pp．131-136．
[2] 刘昌平．汉字识别技术现状与展望．中国中文信息学会二十周年学术会议论文集．2001，pp．108-110．
[3] Song Mao and Tapas Kanungo. Empircal Performance Evaluation of Page Segmentation Algorithms. in Proceeding of SPIE Conference on Document Recognition and Retrieval Ⅶ. 2000, vol. 3967, pp. 303-312.
[4] 夏波涌．文档图像理解研究：[学位论文]．中国科技大学．1998．
[5] Susan E. Hauser, Daniel X. Le, George R. Thoma. Automated Zone Correction in Bitmapped Document image, in Proceeding of SPIE Conference on Document Recognition and Retrieval Ⅶ. 2000, vol. 3967, pp. 238-258.
[6] 张利，朱颖．基于游程平滑的英文版面分割．电子学报．1999，Vol．27，No．7．
[7] Lawrence O' Gorman. The Document Spectrum For Page Layout Analysis. IEEE Transaction on PAMI. 1993, Vol. 15, No. 11, pp. 1162-1173.
[8] 陈明，丁晓青．复杂中文报纸的版面分析、理解和重构．清华大学学报．2001，Vol．41，No．1．
[9] Liody Alan Fletcher and Rangacher kasturl. A Robust Algorithm for Text String Separation from Mixed Text/Graphics Images. IEEE Trans. PAMI. 1988, Vol. 10, No. 6, pp. 910-918.
[10] 周长岭．中文OCR中的版面分析算法初探．第六届全国汉字识别学术会议论文集．1996，pp．137-142．
[11] 马希荣，王行愚．基于汉字字型的西夏文字版面分析的研究．计算机工
程与应用．2001，No．1．
[12] Hiroaki Takebe, Yutaka Katsuyama, and Satoshi Naoi. Character string extraction from newpaper headlines with a background design by recognizing a combination of connected components in Proceeding of SPIE Conference on Document Recognition and Retrieval Ⅵ. 1999, vol. 3561, pp. 22-29.
[13] 刘昊．基于背景描述的中文版面分析方法的研究：[学位论文]．哈尔滨工业大学．1999．
[14] A. Antonacopoulos. Page Segmentation Using the Description of the Background. Computer Vision and Image Understanging. 1998, Vol. 70, No. 3, pp. 350-369.
[15] 田学东，郭宝兰．基于组合特征的中文版面分析．中文信息学报．1999，Vol．13，No．4．
[16] SATOSHI NAOI, FUJITSU LABORATORIES LID. Hierarchical Rectangular Representation for Connected Component Labeled Image. in Proceeding of SPIE Conference on Document Recognition and Retrieval Ⅶ. 2000, vol. 3567, pp. 70-77.
[17] 刘定强，张忻中．基于组件的中文版面分析．中文信息学报．2000，Vol．14，No．2，pp．8—13．
[18] Jiming Liu, Yuan Y. Tang. Adaptive Image Segmentation With Distributed Behavior-Based Agents. IEEE Trans. PAMI. 1999, Vol.21, No. 6, pp. 544-550.
[19] Kamran Etemad 等. Multiscale Document Page Segmentation Using Soft Decision Integration, IEEE Transaction on PAMI,1997, Vol. 19, No 1, pp. 92-96.
[20] James Z. Wang, Jia Li, and Gio Wiederhold. Unsupervised Multilsolution for Images With Low Depth of Field. IEEE Trans. PAMI. 2001, Vol.23, No. 1, pp. 85-90.
[21] Sarour N. Sriharl. Machine Printed Japanese Document Recognition.
Pattern Recognition. 1997, Vol. 30, No. 8, pp. 1301-1313.
[22] Kyong-Ho Lee, Yoon-Chul Choy, and Sung-Bae Cho, Geometric Structure Analysis of Document Images: A knowledge-Based Approach, IEEE Trans. PAMI, 2000, Vol. 22, No. 11, pp. 1224-1240.
[23] 江世盛．中文版面分析：[学位论文]．中科院自动化所．1999．
[24] 张利，朱颖．吴国威．版面分割中文本区域最佳结构表示树生成算法．中国图像图像学报．1998，Vol．3，No．7．
[25] Victor Wu, Raghavan Manmatha, and Edward M. TextFinder:An Automatic System to Detect and Recognize Text In Images, IEEE Trans. PAMI. 1999, Vol. 21, No. 11,pp. 1224-1229.
[26] Stefan Agne, Markus Rogger. Benchmarking of Document Page Segmentation, in Proceeding of SPIE Conference on DocumOnt Recognition and Retrieval Ⅶ. 2000, vol. 3567, pp. 165-170.
[27] LIOYD ALAN LIETCHER, RANGACHAR KASTURI. A Robut Algorithm for Text String Separation from Mixed Text/Graphics Images. IEEE Transaction on PAMI. 1988, Vol.10, No. 6, pp. 910-918.
[28] 胡民，邓振岳，周利华．图像获取系统与处理软件的连接．电子技术与外部设备．1999，Vol．23，No．1．
[29] 周文，卢章平．基于TWAIN技术的图像获取应用程序设计．微型电脑应用．2000，Vol．16，No．5，pp．47-48．
[30] Yi-Kai Chen, Jhing-Fa Wang. Skew detection and reconstruction based on maximization of variance of transition-counts. Pattern Recognition. 33 (2000) 195-208.
[31] 毛经坤，罗予频．基于Hough变换的中文倾斜印刷字倾斜角度检测算法．模式识别与人工智能．2000，Vol．13，No．3．
[32] 姜哲，马少平，夏莹．大型中文古籍《四库全书》自动版面分析系统．中文信息学报．2000，Vol．14，No．2．
[33] 吕岳，施鹏飞，张克华．基于组件合并的手写体汉字串分割．软件学报．2000，Vol．11．No，11．
[34] 郭燕慧，王小捷，钟义信．文本倾向识别的置信度估计．计算机工程与应用．2001，No．1．
[35] 郑苏民，张松顺．一种新的文本预处理方法的研究．第四届全国汉字及语音识别学术会议论文集．1992，pp．44-52．
[36] 宋东．版面分析与印刷汉字识别研究：[学位论文]．中科院计算技术研究所，1999．
[37] 张忻中．汉字识别技术．清华大学出版社．1992．
[38] 胡家忠．计算机文字识别技术．气象出版社．1994．
[2] 刘昌平．汉字识别技术现状与展望．中国中文信息学会二十周年学术会议论文集．2001，pp．108-110．
[3] Song Mao and Tapas Kanungo. Empircal Performance Evaluation of Page Segmentation Algorithms. in Proceeding of SPIE Conference on Document Recognition and Retrieval Ⅶ. 2000, vol. 3967, pp. 303-312.
[4] 夏波涌．文档图像理解研究：[学位论文]．中国科技大学．1998．
[5] Susan E. Hauser, Daniel X. Le, George R. Thoma. Automated Zone Correction in Bitmapped Document image, in Proceeding of SPIE Conference on Document Recognition and Retrieval Ⅶ. 2000, vol. 3967, pp. 238-258.
[6] 张利，朱颖．基于游程平滑的英文版面分割．电子学报．1999，Vol．27，No．7．
[7] Lawrence O' Gorman. The Document Spectrum For Page Layout Analysis. IEEE Transaction on PAMI. 1993, Vol. 15, No. 11, pp. 1162-1173.
[8] 陈明，丁晓青．复杂中文报纸的版面分析、理解和重构．清华大学学报．2001，Vol．41，No．1．
[9] Liody Alan Fletcher and Rangacher kasturl. A Robust Algorithm for Text String Separation from Mixed Text/Graphics Images. IEEE Trans. PAMI. 1988, Vol. 10, No. 6, pp. 910-918.
[10] 周长岭．中文OCR中的版面分析算法初探．第六届全国汉字识别学术会议论文集．1996，pp．137-142．
[11] 马希荣，王行愚．基于汉字字型的西夏文字版面分析的研究．计算机工
程与应用．2001，No．1．
[12] Hiroaki Takebe, Yutaka Katsuyama, and Satoshi Naoi. Character string extraction from newpaper headlines with a background design by recognizing a combination of connected components in Proceeding of SPIE Conference on Document Recognition and Retrieval Ⅵ. 1999, vol. 3561, pp. 22-29.
[13] 刘昊．基于背景描述的中文版面分析方法的研究：[学位论文]．哈尔滨工业大学．1999．
[14] A. Antonacopoulos. Page Segmentation Using the Description of the Background. Computer Vision and Image Understanging. 1998, Vol. 70, No. 3, pp. 350-369.
[15] 田学东，郭宝兰．基于组合特征的中文版面分析．中文信息学报．1999，Vol．13，No．4．
[16] SATOSHI NAOI, FUJITSU LABORATORIES LID. Hierarchical Rectangular Representation for Connected Component Labeled Image. in Proceeding of SPIE Conference on Document Recognition and Retrieval Ⅶ. 2000, vol. 3567, pp. 70-77.
[17] 刘定强，张忻中．基于组件的中文版面分析．中文信息学报．2000，Vol．14，No．2，pp．8—13．
[18] Jiming Liu, Yuan Y. Tang. Adaptive Image Segmentation With Distributed Behavior-Based Agents. IEEE Trans. PAMI. 1999, Vol.21, No. 6, pp. 544-550.
[19] Kamran Etemad 等. Multiscale Document Page Segmentation Using Soft Decision Integration, IEEE Transaction on PAMI,1997, Vol. 19, No 1, pp. 92-96.
[20] James Z. Wang, Jia Li, and Gio Wiederhold. Unsupervised Multilsolution for Images With Low Depth of Field. IEEE Trans. PAMI. 2001, Vol.23, No. 1, pp. 85-90.
[21] Sarour N. Sriharl. Machine Printed Japanese Document Recognition.
Pattern Recognition. 1997, Vol. 30, No. 8, pp. 1301-1313.
[22] Kyong-Ho Lee, Yoon-Chul Choy, and Sung-Bae Cho, Geometric Structure Analysis of Document Images: A knowledge-Based Approach, IEEE Trans. PAMI, 2000, Vol. 22, No. 11, pp. 1224-1240.
[23] 江世盛．中文版面分析：[学位论文]．中科院自动化所．1999．
[24] 张利，朱颖．吴国威．版面分割中文本区域最佳结构表示树生成算法．中国图像图像学报．1998，Vol．3，No．7．
[25] Victor Wu, Raghavan Manmatha, and Edward M. TextFinder:An Automatic System to Detect and Recognize Text In Images, IEEE Trans. PAMI. 1999, Vol. 21, No. 11,pp. 1224-1229.
[26] Stefan Agne, Markus Rogger. Benchmarking of Document Page Segmentation, in Proceeding of SPIE Conference on DocumOnt Recognition and Retrieval Ⅶ. 2000, vol. 3567, pp. 165-170.
[27] LIOYD ALAN LIETCHER, RANGACHAR KASTURI. A Robut Algorithm for Text String Separation from Mixed Text/Graphics Images. IEEE Transaction on PAMI. 1988, Vol.10, No. 6, pp. 910-918.
[28] 胡民，邓振岳，周利华．图像获取系统与处理软件的连接．电子技术与外部设备．1999，Vol．23，No．1．
[29] 周文，卢章平．基于TWAIN技术的图像获取应用程序设计．微型电脑应用．2000，Vol．16，No．5，pp．47-48．
[30] Yi-Kai Chen, Jhing-Fa Wang. Skew detection and reconstruction based on maximization of variance of transition-counts. Pattern Recognition. 33 (2000) 195-208.
[31] 毛经坤，罗予频．基于Hough变换的中文倾斜印刷字倾斜角度检测算法．模式识别与人工智能．2000，Vol．13，No．3．
[32] 姜哲，马少平，夏莹．大型中文古籍《四库全书》自动版面分析系统．中文信息学报．2000，Vol．14，No．2．
[33] 吕岳，施鹏飞，张克华．基于组件合并的手写体汉字串分割．软件学报．2000，Vol．11．No，11．
[34] 郭燕慧，王小捷，钟义信．文本倾向识别的置信度估计．计算机工程与应用．2001，No．1．
[35] 郑苏民，张松顺．一种新的文本预处理方法的研究．第四届全国汉字及语音识别学术会议论文集．1992，pp．44-52．
[36] 宋东．版面分析与印刷汉字识别研究：[学位论文]．中科院计算技术研究所，1999．
[37] 张忻中．汉字识别技术．清华大学出版社．1992．
[38] 胡家忠．计算机文字识别技术．气象出版社．1994．
© 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 |