High Accuracy Handwritten Chinese Character Recognition Based on Support Vector Machine and Independent Component Analysis

Conference paper
Part of the Lecture Notes in Electrical Engineering book series (LNEE, volume 208)

Abstract

This paper proposed a new method for handwritten Chinese character recognition based on a combination of independent component analysis (ICA) and support vector machine (SVM). First, we extracted independent basis images of handwritten Chinese character image and the projection vector by using fast ICA algorithm, and obtained the feature vector. Then, we used two stage classification methods based on SVM for classification. The scheme took full advantage of good extraction local features capability of ICA and strong classification ability of SVM, thus increasing the system’s recognition rate. The experiments show that the feature extraction method based on ICA is superior to that of gradient-based, and the two stage classifiers based on SVM is better than that of modified quadratic discriminant function. On HCL2000, a handwritten Chinese character database, the recognition accuracy of 99.87 % has been achieved.

Keywords

Handwritten chinese character recognition Independent component analysis Support vector machine Feature extraction 

Notes

Acknowledgments

The work was partially supported by Key Projects of Education Department of Sichuan Province under Grant No. 10ZA186.

References

  1. 1.
    He ZG, Cao YD (2008) Survey of offline handwritten chinese character recognition. Comput Eng 34(15):201–204Google Scholar
  2. 2.
    Liu CL, Fujisawa H (2008) Classification and learning methods for character recognition: advances and remaining problems. Stud Comput Intell 90:139–161CrossRefGoogle Scholar
  3. 3.
    Zhu CH, Shi CY, Wang JP et al. (2011) Study of offline handwritten chinese character recognition based on dynamic pruned FSVMs. In: International conference on electrical and control engineering, vol 123. pp 395–398Google Scholar
  4. 4.
    Liu CL (2007) Normalization-cooperated gradient feature extraction for handwritten character recognition. IEEE Trans Pattern Anal Mach Intell 29(8):1465–1469CrossRefGoogle Scholar
  5. 5.
    Rui T, Shen C, Ding J et al (2005) Handwritten digit character recognition by model reconstruction based on independent component analysis. J Comput Aided Des Comput Graph 17(3):455–460Google Scholar
  6. 6.
    Bartlett MS (1998) Face image analysis by unsupervised learning and redundancy reduction, vol 12(31). Dissertation, University of California, San Diego, pp 93–99Google Scholar
  7. 7.
    Dong JX, Krzy_zak A, Suen CY (2005) An improved handwritten chinese character recognition system using support vector machine. Pattern Recogn Lett 26:1849–1856Google Scholar
  8. 8.
    Dai R, Liu CL, Xiao B (2007) Chinese character recognition: history, status and prospects. Frontiers Comput Sci China 1(2):126–136CrossRefGoogle Scholar
  9. 9.
    Hyvärinen A, Oja E (2000) Independent component analysis: algorithms and applications. Neural Netw 13(4):411–430CrossRefGoogle Scholar

Copyright information

© Springer-Verlag London 2013

Authors and Affiliations

  1. 1.School of Computer SciencePanzhihua UniversityPanzhihuaChina

Personalised recommendations