Advertisement

An Oversegmentation Method for Handwritten Character Segmentation

  • Magdalena Brodowska
Conference paper
Part of the Advances in Intelligent and Soft Computing book series (AINSC, volume 95)

Abstract

This paper presents a method for generation of set of potential splitting paths for a word image. It combines some existing oversegmentation techniques with novel approaches based on joint use of word profile extrema and external background analysis and on word image skeleton. It can be used as initial segmentation step in most systems that are able to take advantage of oversegmenation. Experiment conducted for this article shows that this method is able to achieve reasonable compromise between oversegmentation rate and accuracy measured as percentage of correctly detected segmentation points.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Bozekova, M.: Comparison of Handwritings. Diploma thesis, Comenius University, Bratislava, Slovak Republic (2008)Google Scholar
  2. 2.
    Huang, L., Wan, G., Liu, C.: An Improved Parallel Thinning Algorithm. In: Proceedings of the Seventh International Conference on Document Analysis and Recognition (2003)Google Scholar
  3. 3.
    Leedham, C.G., Friday, P.D.: Isolating individual handwritten characters. In: Proc. IEE Colloq. Character Recognition and Applications, London (1989)Google Scholar
  4. 4.
    Liang, Z., Shi, P.: A metasynthetic approach for segmenting handwritten Chinese character strings. Pattern Recognition Letters 26, 1498–1511 (2005)CrossRefGoogle Scholar
  5. 5.
    Lu, Z., Chi, Z., Siu, W., Shi, P.: A background-thinning-based approach for separating and recognizing connected handwritten digit strings. Pattern Recognition 32, 921–933 (1999)CrossRefGoogle Scholar
  6. 6.
    Madhvanath, S., Kim, G., Govindaraju, V.: Chaincode Contour Processing for Handwritten Word Recognition. IEEE Transactions on Pattern Analysis and Machine Intelligence 21(9), 928–932 (1999)CrossRefGoogle Scholar
  7. 7.
    Morita, M., Lethelier, E., Yacoubi, A.E., Bortolozzi, F., Sabourin, R.: An HMM-based Approach for Date Recognition. In: Proceedings of the Fourth IAPR International Workshop on Document Analysis Systems (2000)Google Scholar
  8. 8.
    Nicchiotti, G., Scagliola, C., Rimassa, S.: Simple And Effective Cursive Word Segmentation Method. In: Proceedings of the Seventh International Workshop on Frontiers in Handwriting Recognition (2000)Google Scholar
  9. 9.
    Oliveira, L.S.: Automatic Recognition of Handwritten Numerical Strings. PhD thesis, Ecole de Technologie Superieure, Canada (2003)Google Scholar
  10. 10.
    Shubhangi, D.C., Hiremath, P.S.: Handwritten English Character And Digit Recognition Using Multiclass SVM Classifier And Using Structural Micro Features. International Journal of Recent Trends in Engineering 2, 193–195 (2009)Google Scholar
  11. 11.
    Verma, B.: Contour Code Feature Based Segmentation For Handwriting Recognition. In: Proceedings of the Seventh International Conference on Document Analysis and Recognition (2003)Google Scholar
  12. 12.
    Xiao, X., Leedham, G.: Knowledge-based English cursive script segmentation. Pattern Recognition Letters 21, 945–954 (2000)CrossRefGoogle Scholar
  13. 13.
    Yanikoglu, B., Sandon, P.A.: Segmentation of off-line cursive handwriting using linear programming. Pattern Recognition 31(12), 1825–1833 (1998)CrossRefGoogle Scholar
  14. 14.
    Zhao, S., Chi, Z., Shi, P., Yan, H.: Two-stage segmentation ofunconstrained handwritten Chinese characters. Pattern Recognition 36, 145–156 (2003)CrossRefzbMATHGoogle Scholar
  15. 15.
    Zheng, L., Hassin, A.H., Tang, X.: New algorithm for machine printed Arabic character segmentation. Pattern Recognition Letters 25, 1723–1729 (2004)CrossRefGoogle Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2011

Authors and Affiliations

  • Magdalena Brodowska
    • 1
  1. 1.Faculty of Physics, Astronomy and Applied Computer ScienceJagiellonian UniversityKrakowPoland

Personalised recommendations