Elsevier

Pattern Recognition

Volume 31, Issue 3, March 1998, Pages 235-255
Pattern Recognition

Recognition of handwritten digits based on contour information

https://doi.org/10.1016/S0031-3203(97)00046-0Get rights and content

Abstract

In this paper, a new character recognition method based on topological properties, positions of starting point, statistical analysis, and shape recognition is proposed for recognition of unconstrained handwritten digits based on various contour information. First, input digits are classified into three groups according to their topological properties. In group 1, the features of normalized length of the digit contour, the normalized area of the digit and Fourier descriptors of the outer contour of a digit are used for recognition of digits. In group 2, the relative position of outer and interior contour centroids, the position of interior contour centroid, the mean and the variance of the outer contour distance function, and Fourier descriptors of the outer contour of a digit are used for the recognition of digits. In group 3, usually there is only one digit 8, but because of the complexity of writing styles, some other digits are still classified into this group. The final recognition is based on shape comparison of the input digit with models. In the recognition process, some special models are established and used for recognition of broken digits and digits whose topological properties are destroyed. In our experiment, 1000 digits from the NIST database are used for training and 5278 unseen digits are used for testing. The recognition rate has reached 98.5% with a reliability rate of 99.09%, a substitution rate of 0.91% and a rejection rate of 0.59%.

References (40)

  • T.P. Wallace et al.

    An efficient three-dimensional aircraft recognition algorithm using normalized Fourier descriptors

    Computer Graphics Image Process

    (1980)
  • Z. Shi et al.

    Character image enhancement by selective region-growing

    Pattern Recognition Lett.

    (1996)
  • C.Y. Suen

    Computer Recognition of Unconstrained Handwritten Numerals

  • C.Y. Suen

    Character recognition by computer and applications

  • C.Y. Suen et al.

    Automatic recognition of handprinted characters-The state of the art

  • P. Ahmed et al.

    Computer recognition of totally unconstrained handwritten ZIP codes

    Pattern Recognition and Artificial Intell.

    (1987)
  • F. Ali et al.

    Syntactic recognition of handwritten numerials

    IEEE Trans. Systems Man Cybernet.

    (1987)
  • J. Hu and H. Yan, Structural Decomposition and Description of Printed and Handwritten Characters, Proc. ICPR'96 pp....
  • Y. Mizukami and K. Koga A Handwritten Character Recognition System Using Hierarchical Displacement Extraction...
  • W.T. Chen et al.

    A Hierarchical Deformation Model For On-line Cursive Script Recognition

    Pattern Recognition

    (1994)
  • Cited by (23)

    • Defeating line-noise CAPTCHAs with multiple quadratic snakes

      2013, Computers and Security
      Citation Excerpt :

      Once successful segmentation is achieved, as already mentioned, any of a large variety of recognition algorithms could be used. As examples, recognition rates on the pre-segmented NIST handwritten digit database long ago reached 98.5% using Fourier descriptors computed from the digit contours (Cheng and Yan, 1998) and more recently 99.65% using large neural networks (Ciresan et al., 2010). This term favors anti-parallel normal and gradient vectors, encouraging counterclockwise snakes to shrink around or clockwise snakes to expand to enclose light regions surrounded by dark CAPTCHA characters.1

    • Combination of multiple classifiers using probabilistic dictionary and its application to postcode recognition

      2002, Pattern Recognition
      Citation Excerpt :

      High recognition rate is certainly required for a practical system. On the other hand, high recognition reliability [2,12,16] is crucial too. Maximum correct sorting and minimum mis-sorting are the key requirements of an automatic sorting system.

    • A rotation invariant printed Chinese character recognition system

      2001, Pattern Recognition Letters
      Citation Excerpt :

      It is not a one-to-one mapping. Some undesirable empty slots called “measles” may appear after the rotation operation (Cheng and Yan, 1998). When we discuss whether a function is rotation invariant, it is better to discuss in the continuous image domain.

    • Dynamic tree map for video analysis

      2014, Proceedings of the IEEE International Conference on Industrial Technology
    View all citing articles on Scopus
    View full text