Segmentation of Touching Characters in Formulas

  • Masayuki Okamoto
  • Syougo Sakaguchi
  • Tadashi Suzuki
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 1655)


Segmentation of touching characters in mathematical formulas needs a different method from the one for text lines, because these characters may occur in horizontal, vertical or diagonal directions.Our segmentation method is based on the projection profiles of a given binary image and minimal points of the blurred image obtained by applying the Gaussian kernel to the original image.


  1. 1.
    Okamoto, M., Miao, B.: Recognition of Mathematical Expressions by Using Layout Structure of Symbols. Proc. of ICDAR'91 (1991) 242–250Google Scholar
  2. 2.
    Okamoto, M., Higashi, H.: Structure Analysis and Recognition of Mathematical Expressions. Proc. of ICDAR'95 (1995) 430–437Google Scholar
  3. 3.
    Kahan, S., Pavlidis, T., Baird, H. S.: On the Recognition of Printed Characters of Any Font and Size. IEEE Trans. of PAMI, 9, 2 (1987)Google Scholar
  4. 4.
    Wang, L., Pavlidis, T.: Direct Gray-Scale Extraction of Feature for Character Recognition. IEEE Trans. of PAMI, 15, 10 (1993) 1053–1067Google Scholar
  5. 5.
    Nako, K., Takamatsu, R., Sato, M., Kawarada, H.: A New Character Segmentation Method for Handwritten Documents Based on Differential Structure of Blurred Document Image. Technical Report of IEICE, PRU93-131 (1994) 9–16Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1999

Authors and Affiliations

  • Masayuki Okamoto
    • 1
  • Syougo Sakaguchi
    • 1
  • Tadashi Suzuki
    • 1
  1. 1.Dept. of Information EngineeringShinshu UniversityNaganoJapan

Personalised recommendations