LI Pengchao, PENG Liangrui, WEN Juan, “Rejecting Character Recognition Errors Using CNN Based Confidence Estimation,” Chinese Journal of Electronics, vol. 25, no. 3, pp. 520-526, 2016, doi: 10.1049/cje.2016.05.018
Citation: LI Pengchao, PENG Liangrui, WEN Juan, “Rejecting Character Recognition Errors Using CNN Based Confidence Estimation,” Chinese Journal of Electronics, vol. 25, no. 3, pp. 520-526, 2016, doi: 10.1049/cje.2016.05.018

Rejecting Character Recognition Errors Using CNN Based Confidence Estimation

doi: 10.1049/cje.2016.05.018
Funds:  This work is supported by the National Basic Research Program of China (973 Program) (No.2014CB340506), National Natural Science Foundation of China (No.61261130590, No.61032008), and Tsinghua National Laboratory for Information Science and Technology (TNList) Cross-discipline Foundation.
  • Received Date: 2015-02-13
  • Rev Recd Date: 2015-05-08
  • Publish Date: 2016-05-10
  • Although Optical character recognition (OCR) technology has achieved huge progress in recent years, character misrecognition is inevitable. In order to realize high fidelity content of document digitalization, we propose a new Convolutional neural networks (CNN) based confidence estimationmethod.We detect the misrecognized characters through comparing the confidence value with a preset threshold, so as to leave the recognition errors as embedded images in the output digital documents. We adopted sofmax as the estimation of posteriori probability, overlap pooling and maxout with dropout technologies in CNN architecture design. Experimental results show that our method has achieved an explicit improvement compared to baseline system.
  • loading
  • H. Fujisawa, "A view on the past and Future of Character and Document Recognition" Proc. of ICDAR, Paraná, Brazil, pp.3- 7, 2007.
    Chi Fang, Changsong Liu, Liangrui Peng and Xiaoqing Ding, "Automatic performance evaluation of printed Chinese character recognition systems" IJDAR, Vol.4, No.3, pp.177-182, 2002.
    Ding Xiaoqing, Wen Di, Peng Liangrui and Liu Changsong, "Document digitization technology and its application for digital libraries in China", Proc. of First Workshop on Document Image Analysis for Libraries (DIAL2004), Palo Alto, USA, pp.46-53, 2004.
    M.D. Richard and R.P. Lippmann, "Neural network classifiers estimate Bayesian a posteriori probabilities", Neural Computation, Vol.3, No.4, pp.461-483, 1991.
    Xiaofan Lin, Xiaoqing Ding, Ming Chen, Rui Zhang and Youshou Wu, "Adaptive confidence transform based classifier combination for Chinese character recognition", Pattern Recognition Letters, Vol.19, No.10, pp.975-988, 1998.
    Yoshua Bengio, "Learning deep architectures for AI", Foundations and Trends in Machine Learning, Vol.2, No.1, pp.1-127, 2009.
    G. Hinton and R. Salakhutdinov, "Reducing the dimensionality of data with neural networks", Science, Vol.313, No.5786, pp.504-507, 2006.
    Yann LeCun, Koray Kavukcuoglu and Cl' ement Farabet, "Convolutional networks and applications in vision", Proc. ISCAS 2010, Paris, France, pp.253-256, 2010.
    Y. LeCun, B. Boser, J.S. Denker, D. Henderson, R.E. Howard, W. Hubbard and L.D. Jackel, "Backpropagation applied to handwritten zip code recognition", Neural Computation, Vol.1, No.4, pp.541-551, 1989.
    Y. Simard, Patrice, D. Steinkraus and J.C. Platt, "Best practices for convolutional neural networks applied to visual document analysis", Proc. ICDAR, pp.958-963, 2003.
    Tao Wang, D.J. Wu, Adam Coates and Andrew Y. Ng., "Endto- end text recognition with convolutional neural networks", Proc. of ICPR, pp.3304-3308, 2012.
    Weilin Huang, Yu Qiao and Xiaoou Tang, "Robust scene text detection with convolution neural network induced MSER trees", Proc. of ECCV, pp.497-511, 2014.
    K. Chellapilla, M. Shilman and P. Simard, "Optimally combining a cascade of classifiers", Pro. Document Recognition and Retrieval 13, SPIE-IS&T Electronic Imaging, San Jose, USA, pp.6067, 2006.
    Yann LeCun, Yoshua Bengio and Patrick Haffner, "Gradientbased learning applied to document recognition", Proceeding of IEEE, Vol.86, No.11, pp.2278-2324, 1998.
    Malouf, Robert, "A comparison of algorithms for maximum entropy parameter estimation", Proc. Sixth Conf. on Natural Language Learning (CoNLL), pp.49-55, 2002.
    LIU Hailong DING Xiaoqing, "Handwritten character recognition using semi-tied full covariance gaussian mixture model", Chinese Journal of Electronics, Vol.14, No.4, pp.649-652, 2005.
    Md. Musfiqur Rahman Sazal, Sujan Kumar Biswas, Md. Faijul Amin and Kazuyuki Murase, "Bangla handwritten character recognition using deep belief network", Proceeding of EICT, Khulna, Bangladesh, pp.1-5, 2014.
    Utkarsh Porwal, Yingbo Zhou and Venu Govindaraju, "Handwritten arabic text recognition using deep belief networks", Proc. ICPR, Tsukuba, Japan, pp.302-305, 2012.
    Tao Wang, D.J. Wu, A. Coates and A.Y. Ng, "End-to-end text recognition with convolutional neural networks", ICPR, Tsukuba, Japan, pp.3304-3308, 2012.
    Ian J. Goodfellow, David Warde-Farley, Mehdi Mirza, Aaron Courville and Yoshua Bengio, "Maxout networks", JMLR, arXiv:1302.4389, 2013.
    G.E. Hinton, N. Srivastava, A. Krizhevsky, I. Sutskever and R.R. Salakhutdinov, "Improving neural networks by preventing co-adaptation of feature detectors", arXiv:1207.0580, 2012.
    A. Krizhevsky, I. Sutskever and G. Hinton, "ImageNet classification with deep convolutional neural networks", NIPS, 2012.
    Y. Jia, E. Shelhamer, J. Donahue, S. Karayev, J. Long, R. Girshick, S. Guadarrama and T. Darrell, "Caffe: Convolutional architecture for fast feature embedding", Tech report for the Caffe software, arXiv preprint arXiv:1408.5093, 2014.
    MIAO Fuyou, XIONG Yan, CHEN Huanhuan and WANG Xingfu, "A fuzzy quantum neural network and its application in pattern recognition", Chinese Journal of Electronics, Vol.14, No.3, pp.524-528, 2005.
  • 加载中


    通讯作者: 陈斌,
    • 1. 

      沈阳化工大学材料科学与工程学院 沈阳 110142

    1. 本站搜索
    2. 百度学术搜索
    3. 万方数据库搜索
    4. CNKI搜索

    Article Metrics

    Article views (459) PDF downloads(612) Cited by()
    Proportional views


    DownLoad:  Full-Size Img  PowerPoint