A Layout-Free Method for Extracting Elements from Document Images

  • Tsukasa Kochi
  • Takashi Saitoh
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 1655)


SGML is a language for defining the layout structure of a document. Various attempts at generating SGML from a document image have not been successful. We focus on extracting some of the important layout elements by using flexible matching strategy and easy model generation. Our proposed approach treats each extracted element as it were independent. Some segmented areas like “title” or “author” are defined locally making the system robust, able to withstand shifting and noise. The system is also easy to operate. Since the system is not full automatic, we need to supply typical models of each component. Our GUI presents the attributes of each segmented area as well as the original bit map images. The color-coded attributes help us to easily edit the extracted component. In experiments with 288 pages of test images, the proposed method is shown to be 95.6% correct for a wide range of documents. By using 145 pages of documents as a learning set, the system recognized 99.2% of feature sets from 148 various types of unknown documents.


Document Image Segmented Area Input Document Document Image Analysis Element Extraction 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


  1. 1.
    T. Watanabe, et al, “ Extraction of data from preprinted forms”, IEEE Trans. Pattern Analysis and Machine Intelligence, Vol.17, No. 4, 1995, pp.432–445.CrossRefGoogle Scholar
  2. 2.
    J. Yuan, et al, “ Form items extraction by model matching”, ICPR'96,1996, pp.691–695.Google Scholar
  3. 3.
    H. Arai, K. Odaka, “ Information Acquisition and Storage of Forms in Document Processing”, ICDAR, 1997, pp.164–169.Google Scholar
  4. 4.
    C. Wenzel,” Supporting Information Extraction from Printed Documents by Lexico-Semantic Pattern Matching”, ICDAR, 1997.Google Scholar
  5. 5.
    M. Sharpe, et al, “ An Intelligent Document Understanding & Reproduction System”, MVA'94,1994, pp.267–271.Google Scholar
  6. 6.
    T. Watanabe, X. Huang,” Automatic Acquisition of Layout Knowledge for Understanding Business Cards”, ICDAR, 1997, pp.216–220Google Scholar
  7. 7.
    H. Walischewski, “ Automatic Acquisition of Spatial Document Interpretation”, ICDAR, 1997, pp.243–247Google Scholar
  8. 8.
    C. Lin, et al, “ Logical Structure Analysis of Book Document Images Using Contents Information”, ICDAR, 1997, pp.1048–1054.Google Scholar
  9. 9.
    Y. Tang, et al, “ Document Processing for Automatic Knowledge Acquisition”, IEEE, Transaction on Knowledge and Data Engineering, Vol. 6, No. 1, 1994, pp.3–31.CrossRefGoogle Scholar
  10. 10.
    T. Saitoh,et al, ” Document Image Segmentation and Text Area Ordering” Proceedings of ICDAR, 1993, pp.323–329.Google Scholar
  11. 11.
    S. Khoubyari and J. Hull,” Font Function Word Identification in Document Recognition”, CVIU,1996, pp.66–74.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1999

Authors and Affiliations

  • Tsukasa Kochi
    • 1
  • Takashi Saitoh
    • 1
  1. 1.Information and Communication Research and Development Center 32 Research GroupRICOH COMPANY,LTD.KanagawaJapan

Personalised recommendations