一种在低质量图像上提高字符识别率的深度学习框架
|
摘要
论文为了解决低质量图像给识别任务带来的困难,构造了一个由图像增强网络(EnCNN)和手写体数字识别网络(LeNet-5)组成的低质量图片识别框架。将图像增强网络嫁接在识别网络前,并使用论文提出的策略进行模型学习。使得低质量图像在被识别前图像质量得到较大的改善,最终实现低质量手写体图像识别率的提高。实验部分将论文提出的方法和在单纯使用低质量图像或高清图作为训练集进行训练的方法进行了对比,实验表明在低质量图像上,论文提出的方法有更高的数字识别率,且有更强的泛化能力。
Based on the convolutional neural network model for handwritten numeral identification,this paper constructs an image framework with image enhancement network and a handwritten digital identification network in order to solve the difficulties caused by low quality image. The image enhancement network is grafted to identify the network,and the model proposed by this paper is used to model the learning. Which makes the image quality of the low quality image be improved greatly before the recognition task is carried out,and finally the recognition rate of the low quality handwriting image is improved. In this paper,the method proposed in this paper is compared with the method of enhancing the training set. Experiments show that the proposed method is better to other methods and has stronger robustness.
Based on the convolutional neural network model for handwritten numeral identification,this paper constructs an image framework with image enhancement network and a handwritten digital identification network in order to solve the difficulties caused by low quality image. The image enhancement network is grafted to identify the network,and the model proposed by this paper is used to model the learning. Which makes the image quality of the low quality image be improved greatly before the recognition task is carried out,and finally the recognition rate of the low quality handwriting image is improved. In this paper,the method proposed in this paper is compared with the method of enhancing the training set. Experiments show that the proposed method is better to other methods and has stronger robustness.
引文
[1]张猛,余仲秋,姚绍文.手写体数字识别中图像预处理的研究[J].微计算机信息,2006,22(16):256-258.ZHANG Meng,YU Zhongqiu,YAO Shaowen. Image Pretreatment Research in Recognition of Handwritten Numerals[J]. Microcomputer information,2006,22(16):256-258.
[2]Fujisawa Y,Sawa K,Wakabayashi T,et al. Handwritten Numeral Recognition using Gradient and Curvature of Gray Scale Image[II][J]. Pattern Recognition,2002,35(10):2051-2059.
[3]马瑞.非限制手写字符分割中相关技术与算法的研究[D].南京:南京理工大学,2007.MA Rui,Research on Related Tchniques and Algorithms in Unconstrained Handwritten Character Segmentation[D]. Nanjing:Nanjing University of Science and Technology,2007.
[4]董慧.手写体数字识别中的特征提取和特征选择研究[D].北京:北京邮电大学,2007.DONG Hui. Feature extraction and feature selection in handwritten digit recognition[D]. Beijing:Beijing University of Posts and Telecommunications,2007.
[5]庞东虎,金伟杰.英文字符特征提取系统[J].计算机仿真,2007,24(12):208-210.PANG Donghu,JIN Weijie. English Character Feature Extraction[J]. Computer simulation,2007,24(12):208-210.
[6]焦娜.基于软K段主曲线的LPR字符特征的提取方法[J].计算机科学,2017,44(9):49-52.JIAO Na. Extraction Method of LPR Characters Features Based on Soft K-segments Algorithm for Principal Curves[J]. Computer science,2017,44(9):49-52.
[7]王建平,盛军,朱程辉.基于小波分析的视频图像字符特征提取方法研究[J].微电子学与计算机,2002,19(5):51-53.WANG Jianping,SHENG Jun,ZHU Chenghui. Character Feature Extraction of Video Image Based on Wavelet Analysis[J]. Microelectronics and computer,2002,19(5):51-53.
[8]iLécun Y,Bottou L,Bengio Y,et al. Gradient-based learning applied to document recognition[J]. Proceedings of the IEEE,1998,86(11):2278-2324.
[9] Zeiler M D,Fergus R. Visualizing and Understanding Convolutional Networks[M]. Computer Vision-ECCV2014. Springer International Publishing,2014:818-833.
[10]Bahlmann C,Haasdonk B,Burkhardt H. On-Line Handwriting Recognition with Support Vector Machines—A Kernel Approach[C]//Eighth International Workshop on Frontiers in Handwriting Recognition. IEEE Computer Society,2002:49.
[11]邓蕊,刘尧猛,丁忠林.字符识别中支持向量机抑制噪声能力的分析[J].计算机工程与设计,2007,28(16):3963-3964.DENG Rui,LIU Yaomeng,DING Zhonglin. Analysis of Resistance to Noisy Input of Support Vector Machine in Character Recognition[J]. computer engineering and design,2007,28(16):3963-3964.
[12]Dong C,Chen C L,He K,et al. Learning a Deep Convolutional Network for Image Super-Resolution[J]. 2014,8692:184-199.
[13] Deng J,Dong W,Socher R,et al. ImageNet:A large-scale hierarchical image database[C]//Computer Vision and Pattern Recognition, 2009. CVPR 2009.IEEE Conference on. IEEE,2009:248-255.
[14]Krizhevsky A,Sutskever I,Hinton G E. ImageNet classification with deep convolutional neural networks[C]//International Conference on Neural Information Processing Systems. Curran Associates Inc. 2012:1097-1105.
[15]Tang Y,Wu X. Saliency Detection via Combining Region-Level and Pixel-Level Predictions with CNNs[C]//European Conference on Computer Vision. Springer International Publishing,2016:809-825.
[16]Zhao R,Ouyang W,Li H,et al. Saliency detection by multi-context deep learning[C]//Computer Vision&Pattern Recognition,2015:1265-1274.
[17]Pan H,Jiang H. A Deep Learning Based Fast Image Saliency Detection Algorithm[J]. Computer Vision and Pattern Recognition,2016.
[18]Ren S,He K,Girshick R,et al. Faster R-CNN:Towards Real-Time Object Detection with Region Proposal Networks.[J]. IEEE Transactions on Pattern Analysis&Machine Intelligence,2015,39(6):1137-1149.
[19]Badrinarayanan V,Kendall A,Cipolla R. SegNet:A Deep Convolutional Encoder-Decoder Architecture for Scene Segmentation[J]. IEEE Transactions on Pattern Analysis&Machine Intelligence,2017,PP(99):1-1.
[20]Li J,Cheng J H,Shi J Y,et al. Brief Introduction of Back Propagation(BP)Neural Network Algorithm and Its Improvement[M]. Advances in Computer Science and Information Engineering. Springer Berlin Heidelberg,2012:553-558.
[21]Shen L,Lin Z,Huang Q. Relay Backpropagation for Effective Learning of Deep Convolutional Neural Networks[J]. Computer Science,2015:467-482.
[22]Lecun Y,Cortes C. The mnist database of handwritten digits[J]. 2010.
[2]Fujisawa Y,Sawa K,Wakabayashi T,et al. Handwritten Numeral Recognition using Gradient and Curvature of Gray Scale Image[II][J]. Pattern Recognition,2002,35(10):2051-2059.
[3]马瑞.非限制手写字符分割中相关技术与算法的研究[D].南京:南京理工大学,2007.MA Rui,Research on Related Tchniques and Algorithms in Unconstrained Handwritten Character Segmentation[D]. Nanjing:Nanjing University of Science and Technology,2007.
[4]董慧.手写体数字识别中的特征提取和特征选择研究[D].北京:北京邮电大学,2007.DONG Hui. Feature extraction and feature selection in handwritten digit recognition[D]. Beijing:Beijing University of Posts and Telecommunications,2007.
[5]庞东虎,金伟杰.英文字符特征提取系统[J].计算机仿真,2007,24(12):208-210.PANG Donghu,JIN Weijie. English Character Feature Extraction[J]. Computer simulation,2007,24(12):208-210.
[6]焦娜.基于软K段主曲线的LPR字符特征的提取方法[J].计算机科学,2017,44(9):49-52.JIAO Na. Extraction Method of LPR Characters Features Based on Soft K-segments Algorithm for Principal Curves[J]. Computer science,2017,44(9):49-52.
[7]王建平,盛军,朱程辉.基于小波分析的视频图像字符特征提取方法研究[J].微电子学与计算机,2002,19(5):51-53.WANG Jianping,SHENG Jun,ZHU Chenghui. Character Feature Extraction of Video Image Based on Wavelet Analysis[J]. Microelectronics and computer,2002,19(5):51-53.
[8]iLécun Y,Bottou L,Bengio Y,et al. Gradient-based learning applied to document recognition[J]. Proceedings of the IEEE,1998,86(11):2278-2324.
[9] Zeiler M D,Fergus R. Visualizing and Understanding Convolutional Networks[M]. Computer Vision-ECCV2014. Springer International Publishing,2014:818-833.
[10]Bahlmann C,Haasdonk B,Burkhardt H. On-Line Handwriting Recognition with Support Vector Machines—A Kernel Approach[C]//Eighth International Workshop on Frontiers in Handwriting Recognition. IEEE Computer Society,2002:49.
[11]邓蕊,刘尧猛,丁忠林.字符识别中支持向量机抑制噪声能力的分析[J].计算机工程与设计,2007,28(16):3963-3964.DENG Rui,LIU Yaomeng,DING Zhonglin. Analysis of Resistance to Noisy Input of Support Vector Machine in Character Recognition[J]. computer engineering and design,2007,28(16):3963-3964.
[12]Dong C,Chen C L,He K,et al. Learning a Deep Convolutional Network for Image Super-Resolution[J]. 2014,8692:184-199.
[13] Deng J,Dong W,Socher R,et al. ImageNet:A large-scale hierarchical image database[C]//Computer Vision and Pattern Recognition, 2009. CVPR 2009.IEEE Conference on. IEEE,2009:248-255.
[14]Krizhevsky A,Sutskever I,Hinton G E. ImageNet classification with deep convolutional neural networks[C]//International Conference on Neural Information Processing Systems. Curran Associates Inc. 2012:1097-1105.
[15]Tang Y,Wu X. Saliency Detection via Combining Region-Level and Pixel-Level Predictions with CNNs[C]//European Conference on Computer Vision. Springer International Publishing,2016:809-825.
[16]Zhao R,Ouyang W,Li H,et al. Saliency detection by multi-context deep learning[C]//Computer Vision&Pattern Recognition,2015:1265-1274.
[17]Pan H,Jiang H. A Deep Learning Based Fast Image Saliency Detection Algorithm[J]. Computer Vision and Pattern Recognition,2016.
[18]Ren S,He K,Girshick R,et al. Faster R-CNN:Towards Real-Time Object Detection with Region Proposal Networks.[J]. IEEE Transactions on Pattern Analysis&Machine Intelligence,2015,39(6):1137-1149.
[19]Badrinarayanan V,Kendall A,Cipolla R. SegNet:A Deep Convolutional Encoder-Decoder Architecture for Scene Segmentation[J]. IEEE Transactions on Pattern Analysis&Machine Intelligence,2017,PP(99):1-1.
[20]Li J,Cheng J H,Shi J Y,et al. Brief Introduction of Back Propagation(BP)Neural Network Algorithm and Its Improvement[M]. Advances in Computer Science and Information Engineering. Springer Berlin Heidelberg,2012:553-558.
[21]Shen L,Lin Z,Huang Q. Relay Backpropagation for Effective Learning of Deep Convolutional Neural Networks[J]. Computer Science,2015:467-482.
[22]Lecun Y,Cortes C. The mnist database of handwritten digits[J]. 2010.