Abstract
A hybrid system is developed for on-line recognition of hand-written Korean characters. Each syllable consists of several Korean alphabets and is written in a rectangular box to result in a 2-dimensional compositions of alphabets. Although only 24 alphabets exist, the number of syllables easily exceeds 3000. Therefore, instead of the syllables, the 24 alphabets are used as basic recognition unit, which causes difficult segmentation and recognition problems. Each Korean alphabet has at most 4 strokes, and 4 neural networks are trained separately for alphabets with different number of strokes. To improve the recognition preformance, after neural network classifications, classical pattern matching is also applied to check validity of the decisions made by the neural network classifiers. The hybrid systems is robust on distortion and rotation of the characters.
This is a preview of subscription content, log in via an institution.
Preview
Unable to display preview. Download preview PDF.
References
S.-B. Cho and J.H. Kim, Recognition of large-set printed Hangul(Korean script) by two-stage backpropagation neural classifier, Pattern Recognition 25(11), 1353–1360 (1992)
J.S. Jang, M.W. Kim, C.-D. Lim and Y.-S. Song, Recognition of handwritten Korean characters based on constraint relaxation, Proc. Pacific Rim Int. Conf. on Artificial Intelligence, pp.1209–1215, Seoul, Korea, Sep. (1992)
Y.H. Huh and H.L. Beust, On-line recognition of hand-printed Korean characters, Pattern Recognition 15(6), 445–453 (1982)
Jin-Young Ha, Se-Chang Oh, Jin-Hyung Kim and Young-Bin Kwon, Unconstrained Handwritten Word Recognition with Interconnected Hidden Markov Models, Proce.of3rdlnternational Workshop on Frontiers in Handwriting Recognition, pp. 455–460, Buffalo, USA, May (1993).
Hee-Soen park and Seong-Whan Lee, Off-Line Recognition of Large-Set Handwritten Hangul with Hidden Markov Models, Proc.of3rd International Workshop on Frontiers in Handwriting Recognition, pp.51–61, Buffalo, USA, May (1993).
U.J. Seong, Online Hand-written Hangul Recognition Using Hierarchical Curve Representation, M.S.Dissertion, Dept. of Computer Science, Korea Advanced Institute of Science and Technology (1991)
Yoon-Seon Song, Myung Won Kim, Jong Moon Kim, A Neural Network Approach to Continuous Handwritten Character Recognition without Segmentation, Internation Conference on Neural Information Processing, pp.122–127, Seoul, Korea, Oct. (1994)
Hendrawan, A. C Downton, C. G. Leedham, A Fuzzy approach to handwritten address verification, Proc.of3rd International Workshop on Frontiers in Handwriting Recognition, pp.300–311, Buffalo, USA, May (1993).
Eduard Sackinger, Bernhard E. Boser, Jane Bromley, Yann LeCun and Lawrence D. Jackel, Application of the ANNA Neural Network Chip to High-Speed Character Recognition, IEEE Trans. on Neural Networks 3(3), pp.498–505 (1992).
Richard P. Lippmann, An Introduction to computing with neural nets, IEEE ASSP Magazine, pp.4–22, April (1987).
Author information
Authors and Affiliations
Editor information
Rights and permissions
Copyright information
© 1995 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Cho, JW., Lee, SY., Park, C.H. (1995). On-line handwritten character recognition by a hybrid method based on neural networks and pattern matching. In: Mira, J., Sandoval, F. (eds) From Natural to Artificial Neural Computation. IWANN 1995. Lecture Notes in Computer Science, vol 930. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-59497-3_269
Download citation
DOI: https://doi.org/10.1007/3-540-59497-3_269
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-59497-0
Online ISBN: 978-3-540-49288-7
eBook Packages: Springer Book Archive