Research on Feature Extraction Method for Handwritten Chinese Character Recognition Based on Supervised Independent Component Analysis

Abstract:

Feature extraction is very difficult for handwritten Chinese character because of large Chinese characters set, complex structure and very large shape variations. The recognition rate by currently used feature extraction methods is far from the requirements of the people. For this problem, a new supervised independent component analysis (SICA) algorithm based on J-divergence entropy is proposed for feature extraction, which can measure the difference between different categories. The scheme takes full advantage of good extraction local features capability and powerful capability to handle data with non-Gaussian distribution by ICA, and the extracted feature component and classification can be tightly combined. The experiments show that the feature extraction method based on SICA is superior to that of gradient-based and that of ICA.

• Periodical: Advanced Materials Research (Volumes 774-776)
• Pages: 1636-1641
• DOI: https://doi.org/10.4028/www.scientific.net/AMR.774-776.1636
• Online since: September 2013
• Authors: Ze Min Liu, Zhi Guo He, Yu Dong Cao
• Keywords: Divergence Entropy, Feature Extraction, Handwritten Chinese Character Recognition, Independent Component Analysis (ICA), Supervised Independent Component Analysis

References

[1] Zhiguo He, Yudong Cao: Survey of Offline Handwritten Chinese Character Recognition. Computer Engineering, Vol. 34, 201-204 (2008) (In Chinese).

[2] Chenglin Liu, Hiromichi Fujisawa: Classification and Learning Methods for Character Recognition: Advances and Remaining Problems. Studies in Computational Intelligence, Vol. 90, 139-161 (2008).

DOI: 10.1007/978-3-540-76280-5_6

[3] Chenghui Zhu, Changyu Shi and Jianping Wang: Study of Offline Handwritten Chinese Character Recognition based on Dynamic Pruned FSVMs. In Proc. International Conference on Electrical and Control Engineering, 395-398 (2011).

DOI: 10.1109/iceceng.2011.6057497

[4] Cheng-Lin Liu, Fei Yin, Da-Han Wang, Qiu-Feng Wang: Online and Offline Handwritten Chinese Character Recognition: Benchmarking on New Databases. Pattern Recognition, Vol. 46, 155-162 (2013).

DOI: 10.1016/j.patcog.2012.06.021

[5] Liu Chenglin, Masashi Koga and Hiromichi Fujisawa: Gabor Feature Extraction for Character Recognition: Comparison with Gradient Feature. In Proc. ICDAR, 121-125 (2005).

DOI: 10.1109/icdar.2005.119

[6] Wu Tianlei, Ma Shaoping: Feature Extraction by Hierarchical Overlapped Elastic Meshing for Handwritten Chinese Character Recognition. In Proc. ICDAR, 529-533 (2003).

DOI: 10.1109/icdar.2003.1227721

[7] Liu ChengLin: Normalization-cooperated Gradient Feature Extraction for Handwritten Character Recognition. IEEE Trans. on Pattern Analysis and Machine Intelligence, Vol. 29, 1465-1469 (2007).

DOI: 10.1109/tpami.2007.1090

[8] Rui Ting, Shen Chunlin, Ding Jian: Handwritten Digit Character Recognition by Model Reconstruction Based on Independent Component Analysis. Journal of Computer Aided Design & Computer Graphics, Vol. 17, 455-460 (2005).

[9] M. S. Bartlett: Face Image Analysis by Unsupervised Learning and Redundancy Reduction. Ph.D. Dissertation, University of California, San Diego (1998).

[10] A. Hyvärinen, E. Oja: Independent Component Analysis: Algorithms and Applications. Neural Networks, Vol. 13, 411-430 (2000).

DOI: 10.1016/s0893-6080(00)00026-5

[11] Dai Ruwei, Liu Chenglin, Xiao Baihua: Chinese Character Recognition: History, Status and Prospects. Frontiers of Computer Science in China, Vol. 1, 126-136 (2007).

DOI: 10.1007/s11704-007-0012-5