• DocumentCode
    1968772
  • Title

    An Improved Analysis Approach of Overbrace/Underbrace Structure in Printed Mathematical Expressions

  • Author

    Li, Zhuo-lun ; Tian, Xue-dong

  • Author_Institution
    Key Lab. in Machine Learning & Comput. Intell., Hebei Univ., Baoding, China
  • fYear
    2010
  • fDate
    30-31 Jan. 2010
  • Firstpage
    58
  • Lastpage
    61
  • Abstract
    In printed mathematical expression recognition, structural analysis is a key step which determines the overall recognition results. As a special structure in expressions, overbrace/underbrace structure becomes a major difficulty of structural analysis as they´re difficult to be distinguished from other structures. On the basis of baseline analysis method of expression structures, this paper presents an improved approach of overbrace/underbrace structure combining region location with variable thresholds. This approach introduces overrun threshold of overbrace/underbrace symbol, which is combined with other features to establish the analysis rules and optimized by testing on a variety of overbrace/underbrace structures. The experimental results show that the approach is effective in determining and analyzing overbrace/underbrace structure.
  • Keywords
    pattern recognition; baseline analysis method; improved analysis approach; overbrace-underbrace structure analysis; overbrace-underbrace symbol; printed mathematical expression recognition; Computational intelligence; Computer science; Educational institutions; Information analysis; Information technology; Machine learning; Mathematics; Oceans; Testing; Underwater communication; baseline; overbrace/underbrace structure; overrun threshold; printed mathematical expression recognition; structural analysis;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Innovative Computing & Communication, 2010 Intl Conf on and Information Technology & Ocean Engineering, 2010 Asia-Pacific Conf on (CICC-ITOE)
  • Conference_Location
    Macao
  • Print_ISBN
    978-1-4244-5634-5
  • Electronic_ISBN
    978-1-4244-5635-2
  • Type

    conf

  • DOI
    10.1109/CICC-ITOE.2010.22
  • Filename
    5439291