一种多模型超图用于手写汉字识别算法
|
摘要
随着银行业提出手填票据自动化处理需求后,对手写汉字的识别技术研究推向新的高潮。由于手写汉字形体复杂多样、训练样本不多,从而导致识别率难以提高。设计一种多模型的超图学习算法来识别手写汉字块,根据训练样本间距离关系构建样本关系阵;以样本的稀疏表示参数为样本间的关系紧密性权重构建另一个样本关系阵;以样本约束法则为基础,以标记样本间的关系权重构建标记样本间的关系阵,融合这几个关系矩阵成为多模型的超图学习框架。通过迭代学习,找出最优的手写汉字块类别归属,在手写汉字块的实验中表现出一定的优势。
With the requirement of automatic handwritten bill processing put forward by the banking industry, the handwritten Chinese character recognition technology has reached a new climax. However, due to the complexity and diversity of handwritten Chinese characters and the lack of training samples, it is difficult to improve the recognition rate. For this reason, this paper designed a multi-model hypergraph learning algorithm to recognize handwritten Chinese character blocks. We constructed a sample relationship matrix according to the distance relationship between training samples. Then another sample relationship matrix was constructed based on the sparse representation of samples as the weight of the relationship between samples. Based on the principle of sample constraints, a relationship matrix between marked samples was constructed by the relationship weights between marked samples. These relationship matrices were fused to a multi-model hypergraph learning framework. By iterative learning, the optimal classification of handwritten Chinese character blocks can be found and the proposed method shows certain advantages in the experiment of handwritten Chinese character blocks.
With the requirement of automatic handwritten bill processing put forward by the banking industry, the handwritten Chinese character recognition technology has reached a new climax. However, due to the complexity and diversity of handwritten Chinese characters and the lack of training samples, it is difficult to improve the recognition rate. For this reason, this paper designed a multi-model hypergraph learning algorithm to recognize handwritten Chinese character blocks. We constructed a sample relationship matrix according to the distance relationship between training samples. Then another sample relationship matrix was constructed based on the sparse representation of samples as the weight of the relationship between samples. Based on the principle of sample constraints, a relationship matrix between marked samples was constructed by the relationship weights between marked samples. These relationship matrices were fused to a multi-model hypergraph learning framework. By iterative learning, the optimal classification of handwritten Chinese character blocks can be found and the proposed method shows certain advantages in the experiment of handwritten Chinese character blocks.
引文
[1] Suen C Y,Xu Q,Lam L.Automatic recognition of handwritten data on Cheques-factor fiction[J].Patten Recognition Letters,1999(20):1287-1295.
[2] Gorski N,Anisimov V,Augustin E,et al.A2iA check reader:a family of bank recognition systems[C]//Fifth International Conference on Document Analysis and Recognition,Bangalore,India.1999:523-526.
[3] 刘斌,肖惠勇.基于不可分小波变换与Zernike矩的印刷体汉字识别方法[J].计算机应用与软件,2018,35(4):227-236.
[4] 朱利娟,云中华,边巴旺堆.基于极坐标变换的脱机手写藏文字符特征提取方法[J].计算机应用与软件,2018,35(3):162-166.
[5] Ye X,Cheriet M,Suen C Y.Stroke-model-based character extraction from gray-level document images[J].IEEE Transactions on Image Processing,2001,10(8):1152-1161.
[6] 季婧婧,娄震.基于二次分割的银行票据彩色印章的滤除[J].现代电子技术,2014,37(22):5-9.
[7] 李洁.票据自动处理相关技术研究及应用[D].西安:西安建筑科技大学,2014.
[8] 杜刚.银行票据识别系统的研究[D].锦州:辽宁工业大学,2016.
[9] 周星辰.基于深度模型的脱机手写体汉字识别研究[D].杭州:浙江大学,2016.
[10] 金连文,钟卓耀,杨钊,等.深度学习在手写汉字识别中的应用综述[J].自动化学报,2016,42(8):1125-1141.
[11] 郭志民,袁少光,孙玉宝.基于L0稀疏超图半监督学习的异常用电行为识别[J].计算机应用与软件,2018,35(2):54-59.
[12] Zhou D,Huang J,Sch?lkopf,Bernhard.Learning with Hypergraphs:Clustering,Classification,and Embedding[C]//Proceedings of the 19th International Conference on Neural Information Processing Systems.MIT Press,2006:1601-1608.
[13] 余婵娟,方昱春.稀疏表示的特征选择方法在人脸性别和种族识别中的应用[J].计算机应用与软件,2016,33(1):138-141,155.
[14] Wright J,Yang A,Ganesh A,et al.Robust face recognition via sparse representation[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2009,31(2):210-227.
[15] Candes E J,Tao T.Decoding by Linear Programming[J].IEEE Transactions on Information Theory,2005,51(12):1-22.
[16] Yu J,Tao D,Rui Y,et al.Pairwise constraints based multiview features fusion for scene classification[J].Pattern Recognition,2013,46(2):483-496.
[17] Zhang T,Tao D,Li X,et al.Patch Alignment for Dimensionality Reduction[J].IEEE Transactions on Knowledge and Data Engineering,2009,21(9):1299-1313.
[18] Bezdek J C,Hathaway R J.Some notes on alternating optimization.[J].Lecture Notes in Computer Science,2002,2275(4):288-300.
[19] 芦兵,孙俊,许晓东.基于正则表达式的图像目标特征提取方法研究[J].计算机应用与软件,2018,35(4):260-264,292.
[20] 周忠义,吴谨,朱磊.基于多路特征融合和深度学习的露霜图像分类[J].计算机应用与软件,2018,35(10):205-210,231.
[21] Chen W S,Dai X,Pan B,et al.Semi-Supervised Discriminant Analysis Method for Face Recognition[J].International Journal of Wavelets Multiresolution and Information Processing,2015,13(6):2979-2984.
[22] Yu J,Tao D,Wang M.Adaptive Hypergraph Learning and its Application in Image Classification[J].IEEE Transactions on Image Processing,2012,21(7):3262-3272.
[2] Gorski N,Anisimov V,Augustin E,et al.A2iA check reader:a family of bank recognition systems[C]//Fifth International Conference on Document Analysis and Recognition,Bangalore,India.1999:523-526.
[3] 刘斌,肖惠勇.基于不可分小波变换与Zernike矩的印刷体汉字识别方法[J].计算机应用与软件,2018,35(4):227-236.
[4] 朱利娟,云中华,边巴旺堆.基于极坐标变换的脱机手写藏文字符特征提取方法[J].计算机应用与软件,2018,35(3):162-166.
[5] Ye X,Cheriet M,Suen C Y.Stroke-model-based character extraction from gray-level document images[J].IEEE Transactions on Image Processing,2001,10(8):1152-1161.
[6] 季婧婧,娄震.基于二次分割的银行票据彩色印章的滤除[J].现代电子技术,2014,37(22):5-9.
[7] 李洁.票据自动处理相关技术研究及应用[D].西安:西安建筑科技大学,2014.
[8] 杜刚.银行票据识别系统的研究[D].锦州:辽宁工业大学,2016.
[9] 周星辰.基于深度模型的脱机手写体汉字识别研究[D].杭州:浙江大学,2016.
[10] 金连文,钟卓耀,杨钊,等.深度学习在手写汉字识别中的应用综述[J].自动化学报,2016,42(8):1125-1141.
[11] 郭志民,袁少光,孙玉宝.基于L0稀疏超图半监督学习的异常用电行为识别[J].计算机应用与软件,2018,35(2):54-59.
[12] Zhou D,Huang J,Sch?lkopf,Bernhard.Learning with Hypergraphs:Clustering,Classification,and Embedding[C]//Proceedings of the 19th International Conference on Neural Information Processing Systems.MIT Press,2006:1601-1608.
[13] 余婵娟,方昱春.稀疏表示的特征选择方法在人脸性别和种族识别中的应用[J].计算机应用与软件,2016,33(1):138-141,155.
[14] Wright J,Yang A,Ganesh A,et al.Robust face recognition via sparse representation[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2009,31(2):210-227.
[15] Candes E J,Tao T.Decoding by Linear Programming[J].IEEE Transactions on Information Theory,2005,51(12):1-22.
[16] Yu J,Tao D,Rui Y,et al.Pairwise constraints based multiview features fusion for scene classification[J].Pattern Recognition,2013,46(2):483-496.
[17] Zhang T,Tao D,Li X,et al.Patch Alignment for Dimensionality Reduction[J].IEEE Transactions on Knowledge and Data Engineering,2009,21(9):1299-1313.
[18] Bezdek J C,Hathaway R J.Some notes on alternating optimization.[J].Lecture Notes in Computer Science,2002,2275(4):288-300.
[19] 芦兵,孙俊,许晓东.基于正则表达式的图像目标特征提取方法研究[J].计算机应用与软件,2018,35(4):260-264,292.
[20] 周忠义,吴谨,朱磊.基于多路特征融合和深度学习的露霜图像分类[J].计算机应用与软件,2018,35(10):205-210,231.
[21] Chen W S,Dai X,Pan B,et al.Semi-Supervised Discriminant Analysis Method for Face Recognition[J].International Journal of Wavelets Multiresolution and Information Processing,2015,13(6):2979-2984.
[22] Yu J,Tao D,Wang M.Adaptive Hypergraph Learning and its Application in Image Classification[J].IEEE Transactions on Image Processing,2012,21(7):3262-3272.
© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号 地址：北京市海淀区学院路29号 邮编：100083 电话：办公室：(+86 10)66554848；文献借阅、咨询服务、科技查新：66554700 |