Handwritten Hangul recognition using deep convolutional neural networks

Original Paper

Abstract

In spite of the advances in recognition technology, handwritten Hangul recognition (HHR) remains largely unsolved due to the presence of many confusing characters and excessive cursiveness in Hangul handwritings. Even the best existing recognizers do not lead to satisfactory performance for practical applications and have much lower performance than those developed for Chinese or alphanumeric characters. To improve the performance of HHR, here we developed a new type of recognizers based on deep neural networks (DNNs). DNN has recently shown excellent performance in many pattern recognition and machine learning problems, but have not been attempted for HHR. We built our Hangul recognizers based on deep convolutional neural networks and proposed several novel techniques to improve the performance and training speed of the networks. We systematically evaluated the performance of our recognizers on two public Hangul image databases, SERI95a and PE92. Using our framework, we achieved a recognition rate of 95.96 % on SERI95a and 92.92 % on PE92. Compared with the previous best records of 93.71 % on SERI95a and 87.70 % on PE92, our results yielded improvements of 2.25 and 5.22 %, respectively. These improvements lead to error reduction rates of 35.71 % on SERI95a and 42.44 % on PE92, relative to the previous lowest error rates. Such improvement fills a significant portion of the large gap between practical requirement and the actual performance of Hangul recognizers.

Keywords

Handwritten Hangul recognition Character recognition  Deep convolutional neural network  Deep learning Gradient-based learning 

References

  1. 1.
    Park, G.-R., Kim, I.-J., Liu, C.-L.: An evaluation of statistical methods in handwritten Hangul recognition. Int. J. Doc. Anal. Recognit. 16(3), 273–283 (2013)CrossRefGoogle Scholar
  2. 2.
    Kang, K.-W., Kim, J.H.: Utilization of hierarchical, stochastic relationship modeling for Hangul character recognition. IEEE Trans. Pattern Anal. Mach. Intell. 26(9), 1185–1196 (2004)CrossRefMathSciNetGoogle Scholar
  3. 3.
    Bengio, Y.: Learning deep architectures for AI. Found. Trends Mach. Learn. 2(1), 1–127 (2009)CrossRefMATHMathSciNetGoogle Scholar
  4. 4.
    Liu, C.-L., Yin, F., Wang, Q.-F., Wang, D.-H.: ICDAR 2011 Chinese handwriting recognition competition (2011). http://www.nlpr.ia.ac.cn/events/HRcompetition/Report.html
  5. 5.
    Cireşan, D.C., Meier, U., Schmidhuber, J.: Multi-column deep neural networks for image classification. In: IEEE Conference on Computer Vision and Pattern Recognition (2012)Google Scholar
  6. 6.
    Cireşan, D.C., Schmidhuber, J.: Multi-column Deep Neural Networks for Offline Handwritten Chinese Character Classification. IDSIA Technical Report No. IDSIA-05-13 (2013)Google Scholar
  7. 7.
    ImageNet Large Scale Visual Recognition Challenge 2013 (ILSVRC2013) Result. http://www.image-net.org/challenges/LSVRC/2013/results.php
  8. 8.
    Kim, H.Y., Kim, J.H.: Hierarchical random graph representation of handwritten characters and its application to Hangul recognition. Pattern Recognit. 34(2), 187–201 (2001)CrossRefMATHGoogle Scholar
  9. 9.
    Jang, S.I.: Post-Processing of Handwritten Hangul Recognition Using Pair-Wise Grapheme Discrimination, Master Thesis, KAIST (2002)Google Scholar
  10. 10.
    Kim, D.-I., Lee, S.-W.: Automatic evaluation of handwriting qualities of handwritten hangul image database, KU-1. In: Proceedings of the 6th IWFHR, pp. 455–464, Taejon, Korea (1998)Google Scholar
  11. 11.
    Kim, D.H., Hwang, Y.S., Park, S.T., Kim, E.J., Paek, S.H., Bang, S.Y.: Handwritten Korean character image database PE92. In: Proceedings of the 2nd ICDAR, pp. 470–473 (1993)Google Scholar
  12. 12.
    Bae, H.J., Yun, J.M., Cha, E.Y.: Neural network for hand-written character recognition using dynamic bar method. In: Proceedings of the Korea Information Science Autumn Conference, vol. 17, issue 2, pp. 251–254 (1990)Google Scholar
  13. 13.
    Kim, M.W., Jang, J.S., Lim, C.D., Song, Y.S., Kim, J.H.: Improvements to a Hierarchical Interaction Neural Network for Context-Dependent Pattern Recognition and Its Experimentation with Handwritten Korean Character Recognition. Technical Report, Electronics and Telecommunication Research Institute, Taejon, Korea (1992)Google Scholar
  14. 14.
    Jeong, S.H.: Handwritten Hangul Recognition Based on Character Cluster Segmentation, Technical Memo. Electronics and Telecommunication Research Institute, Taejon, Korea (2002)Google Scholar
  15. 15.
    Torralba, A., Fergus, R., Freeman, W.: 80 million tiny images: a large dataset for non-parametric object and scene recognition. IEEE Trans. Pattern Anal. Mach. Intell. 30(11), 1958–1970 (2008)CrossRefGoogle Scholar
  16. 16.
    Hinton, G.E., Osindero, S., Teh, Y.: A fast learning algorithm for deep belief nets. Neural Comput. 18(7), 1527–1554 (2006)CrossRefMATHMathSciNetGoogle Scholar
  17. 17.
    Hinton, G.E., Dayan, P., Frey, B.J., Neal, R.: The wake-sleep algorithm for self-organizing neural networks. Science 268, 1158–1161 (1995)CrossRefGoogle Scholar
  18. 18.
    Vincent, P., Larochelle, H., Bengio, Y., Manzagol, P.-A.: Extracting and composing robust features with denoising autoencoders. In: Cohen, W.W., McCallum, A., Roweis, S.T. (eds.) Proceedings of the Twenty-Fifth International Conference on Machine Learning (ICML’08), pp. 1096–1103, ACM (2008)Google Scholar
  19. 19.
    Fukushima, K.: Neocognitron: a self-organizing neural network model for a mechanism of pattern recognition unaffected by shift in position. Biol. Cybern. 36(4), 193–202 (1980)CrossRefMATHGoogle Scholar
  20. 20.
    LeCun, Y., Bottou, L., Bengio, Y., Haffner, P.: Gradient-based learning applied to document recognition. Proc. IEEE 86(11), 2278–2324 (1998)CrossRefGoogle Scholar
  21. 21.
    Simard, D., Steinkraus, P.Y., Platt, J.C.: Best practices for convolutional neural networks. In: Proceedings of the International Conference on Document Analysis and Recognition (ICDAR’03), p. 958. IEEE Computer Society, Washington, DC, USA (2003) Google Scholar
  22. 22.
    Boureau, Y., Ponce, J., LeCun, Y.: A theoretical analysis of feature pooling in vision algorithms. In: Proceedings of the International Conference on Machine Learning (ICML’10) (2010)Google Scholar
  23. 23.
  24. 24.
    Ozkan, C., Erbek, F.: A comparison of activation functions for multispectral Landsat TM image classification. Photogramm. Eng. Remote Sens. 69(11), 1225–1234 (2003)CrossRefGoogle Scholar
  25. 25.
  26. 26.
    Nair, V., Hinton, G.: Rectified linear units improve restricted boltzmann machines. In: Proceedings of the 27th International Conference on Machine Learning (ICML-10) (2010)Google Scholar
  27. 27.
    Glorot, X., Antoine, B., Bengio, Y.: Deep Sparse Rectifier Networks. In: Proceedings of the 14th International Conference on Artificial Intelligence and Statistics. JMLR W&CP, vol. 15 (2011)Google Scholar
  28. 28.
    Goodhill, G., Barrow, H.: The role of weight normalization in competitive learning. Neural Comput. 6(2), 255–269 (1994)CrossRefGoogle Scholar
  29. 29.
    Liu, H., Ding, X.: Handwritten character recognition using gradient feature and quadratic classifier with multiple discrimination schemes. In: Proceedings. 8th International Conference on Document Analysis and Recognition, IEEE (2005)Google Scholar
  30. 30.
    Cireşan, D., et al.: Deep Big Multilayer Perceptrons for Digit Recognition, Neural Networks: Tricks of the Trade, pp. 581--598. Springer, Berlin Heidelberg (2012)Google Scholar
  31. 31.
  32. 32.
    Everitt, B.: The Analysis of Contingency Tables, 2nd edn. Chapman and Hall/CRC, London (1992)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2014

Authors and Affiliations

  1. 1.School of CSEEHandong Global UniversityPohangRepublic of Korea
  2. 2.Department of Computer Science, School of Information and Computer ScienceUniversity of CaliforniaIrvineUSA

Personalised recommendations