基于混合核WLS-SVR的古汉字识别

doi:10.3969/j.issn.0253-2778.2015.04.010

中国科学技术大学学报 ›› 2015, Vol. 45 ›› Issue (4): 321-328.DOI: 10.3969/j.issn.0253-2778.2015.04.010

• 论著 • 上一篇

基于混合核WLS-SVR的古汉字识别

胡根生，孙莹莹，徐玲英，梁栋，孙小棋

1.安徽大学电子信息工程学院，安徽合肥 230601;2.安徽大学学报编辑部，安徽合肥 230039

收稿日期:2014-06-10 修回日期:2014-12-29 接受日期:2014-12-29 出版日期:2014-12-29 发布日期:2014-12-29

Recognition of ancient Chinese characters based on hybrid kernel WLS-SVR

HU Gensheng, SUN Yingying, XU Lingying, LIANG Dong, SUN Xiaoqi

1.School of Electronics and Information Engineering, Anhui University, Hefei 230601, China；2. Editorial Department of Anhui University, Hefei 230039, China

Received:2014-06-10 Revised:2014-12-29 Accepted:2014-12-29 Online:2014-12-29 Published:2014-12-29
Contact: Hu Gensheng
About author:Hu Gensheng(corresponding author), male, born in 1971. PhD/ associate professor. Research field: Machine learning, remote sensing image processing and intelligent algorithm.E-mail：hugs2906@sina.com
Supported by:
Supported by the National Natural Science Foundation of China (61172127), Natural Science Foundation of Anhui Province (1408085MF121).

摘要/Abstract

摘要： 针对现有多种分类器对具有不确定字形的古汉字识别精度不高的问题，提出了一种基于混合核加权最小二乘支持向量回归（WLS-SVR）的古汉字识别算法.WLS-SVR的权重系数采用预测误差的指数衰减函数，混合核是由具有良好局域特性的小波核函数与具有良好全局特性的RBF核函数构成.在特征提取阶段，由于全局点密度与部件结构具有全局特征，而伪二维弹性网格与局部点密度具有局部特征，因此融合了古汉字的全局和局部特征.仿真实验表明，该算法具有较高的准确率与良好的鲁棒性.

关键词: 古汉字识别, WLS-SVR, 混合核, 特征融合

Abstract: The shapes of ancient Chinese characters are often uncertain, which reduces the accuracy of recognition by many classifiers. To solve this problem, a new recognition algorithm combining adaptive weighted least squares support vector regression(WLS-SVR) with hybrid kernel function was proposed to recognize ancient Chinese characters. The weight coefficients of WLS-SVR decayed at a rate of the exponential function of prediction errors. The hybrid kernel was constructed using the wavelet kernel function with local properties and RBF kernel function with global properties. For feature extraction, global point density and component structure are fused with local features of pseudo 2D elastic mesh and local point density. Experiment results show the good robustness and high recognition accuracy of the proposed method.

Key words: ancient Chinese characters recognition, WLS-SVR, hybrid kernel, feature fusion

中图分类号:

TP18

胡根生，孙莹莹，徐玲英，梁栋，孙小棋. 基于混合核WLS-SVR的古汉字识别[J]. 中国科学技术大学学报, 2015, 45(4): 321-328.

HU Gensheng, SUN Yingying, XU Lingying, LIANG Dong, SUN Xiaoqi. Recognition of ancient Chinese characters based on hybrid kernel WLS-SVR[J]. Journal of University of Science and Technology of China, 2015, 45(4): 321-328.

参考文献

［1］
Zhang P. Research of digital construction of Paper files and relics [J]. Cultural Relics of Central Plains, 2009(5):104-107.
[2] Lv X Q, Li M N et al. An oracle classification method based on figure recognition [J]. Journal of Beijing University of Information Science and Technology, 2010(25):92-96.
[3] Chen D, Li N, Li L. Online handwriting recognition research of ancient character[J]. Journal of Beijing Institute of Mechanical Industry, 2008(4):32-37.
[4] Zang G Q. Experiment and Improvement of accuracy of OCR for text-digital image [J]. Intelligence, Information and Sharing, 2010(3):62-67.
[5] Vapnik V. The Nature of Statistical Learning Theory[M]. New York: Springer-Verlag, 1995.
[6] Suykens J A K, Vandewalle J. Least Squares Support Vector Machine Classifiers[J]. Neural Process Letters, 1999 (3):293-300.
[7] Miranian A, Abdollahzade M. Developing a Local Least-Squares Support Vector Machines-Based Neuro-Fuzzy Model for Nonlinear and Chaotic Time Series Prediction[J]. IEEE Transactions on Neural Networks and Learning Systems, 2013, 24(2): 207-218.
[8] Wang L G , Liu D F , Wang Q M et al. Spectral Unmixing Model Based on Least Squares Support Vector Machine With Unmixing Residue Constraints[J]. IEEE Geoscience and Remote Sensing Letters, 2013, 10(6): 1592-1596.
[9] Liu B Y, Yang R G. A novel method based on PCA and LS-SVM for power load forecasting[C]. International Conference on Electric Utility Deregulation and Restructuring and Power Technologies, NanJing, 2008: 759-763.
[10] Zhang H R, Wang X D, Zhang C J et al. Soft sensor technique using LS-SVM and standard SVM[J]. IEEE International Conference on Information Acquisition, Hong Kong and Macau, 2005: 124-127.
[11] Xie J H. Printed character recognition using Kernel CCA with LS-SVM method[C]. Computer and Automation Engineering, 2010: 284-287.
[12] Yin D Y, Wu Y Q. Detection of Small Target in Infrared Image Based on KFCM and LS-SVM[C].International Conference on Intelligent Human-Machine Systems and Cybernetics, 2010: 309-312.
[13] Suykens J A K, de Brabanter J, Lukas L, Vandeewalle J. Weighted Least squares support vector machines: robustness and sparse approximation[J]. Neurocomputing, 2002(48):85-105.
[14] Smits G F, Jordaan E M. Improved SVM regression using mixtures of kernels[C]. textitin Neural Networks, International Joint Conference on, 2002, 3: 2785-2790.
[15] 温昌兵. 基于特征融合的脱机手写体汉字识别[D]. 北京，北京科技大学，2005.
[16] Zhang X B, Huang H, Zhang S J. A FCM clustering algorithm based on Semi-supervised and Point Density Weighted[C]. Intelligent Computing and Intelligent Systems, 2010: 710-713.
[17] Tu Y K, Chen Q H, Huang L. Handwritten Chinese character recognition based on pseudo two-dimensional elastic mesh [J]. Journal of Huazhong University of Science and Technology, 2010(38):38-40.

[1]	马玉莲, 崔文泉. 一种预测高频价格的端到端双目标多任务方法[J]. 中国科学技术大学学报, 2021, 51(3): 246-258.
[2]	陶陶，柏建树，刘恒，侯书东，郑啸. 基于WGAN反馈的深度学习差分隐私保护方法[J]. 中国科学技术大学学报, 2020, 50(8): 1064-1071.
[3]	高翔，陈力. 分组随机梯度下降法：掉队和延迟的平衡[J]. 中国科学技术大学学报, 2020, 50(8): 1156-1161.
[4]	李鹏，郑宇，张谈贵. 基于视觉显著性的无人机目标跟踪[J]. 中国科学技术大学学报, 2020, 50(8): 1162-1169.
[5]	王悦，李京. 基于可视化的卷积神经网络优化方法研究[J]. 中国科学技术大学学报, 2020, 50(7): 959-967.
[6]	崔文泉，余厚莹，侯晓天. 不均衡数据情形的基于聚焦损失的CGAN的集成分类方法[J]. 中国科学技术大学学报, 2020, 50(7): 968-976.
[7]	张怿恺，彭勇，孔万增，文益民. 图正则化的模糊局部坐标编码概念分解模型[J]. 中国科学技术大学学报, 2020, 50(7): 993-1002.
[8]	苏守宝，陈秋鑫，王池社，李智. 群活性反馈的变异自适应分数阶粒子群优化[J]. 中国科学技术大学学报, 2020, 50(7): 1026-1034.
[9]	石陆魁，郭林林，房子哲，张军. 基于Spark的并行ISOMAP算法[J]. 中国科学技术大学学报, 2019, 49(10): 842-850.
[10]	张会敏，杨明，吕静. 基于自适应核联合稀疏表示的多特征高光谱图像分类[J]. 中国科学技术大学学报, 2018, 48(4): 298-306.
[11]	饶齐，杨燕*，滕飞. 基于多视图加权聚类集成的高速列车工况识别[J]. 中国科学技术大学学报, 2018, 48(1): 35-41.
[12]	张皓，吴建鑫，. 集成最大汇合: 最大汇合时只有最大值有用吗[J]. 中国科学技术大学学报, 2017, 47(10): 799-807.
[13]	谷雨，权良虎，陈孟妮，任福继. 基于WiFi背景噪音的被动式人体行为识别研究[J]. 中国科学技术大学学报, 2015, 45(4): 308-313.

基于混合核WLS-SVR的古汉字识别

Recognition of ancient Chinese characters based on hybrid kernel WLS-SVR

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 13

编辑推荐

Metrics

本文评价