计算机工程与应用 ›› 2011, Vol. 47 ›› Issue (24): 150-153.

• 图形、图像、模式识别 • 上一篇    下一篇

抗混叠轮廓波变换的脱线中文手写体笔迹识别

朱贝贝1,尚赵伟1,2,袁 博1,国 庆1,张 峰1,杨建伟3   

  1. 1.重庆大学 计算机学院,重庆 400030
    2.四川省模式识别与智能信息处理重点实验室,成都 610106
    3.南京信息工程大学 数理学院,南京 210044
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2011-08-21 发布日期:2011-08-21

Offline Chinese handwriting-based writer identification with non-aliasing Contourlet transform

ZHU Beibei1,SHANG Zhaowei1,2,YUAN Bo1,GUO Qing1,ZHANG Feng1,YANG Jianwei3   

  1. 1.College of Computer Science,University of Chongqing,Chongqing 400030,China
    2.Key Laboratory of Pattern Recognition and Intelligent Information Processing Sichuan Province,Chengdu 610106,China
    3.College of Mathematics and Physics,Nanjing University of Information Engineering,Nanjing 210044,China
  • Received:1900-01-01 Revised:1900-01-01 Online:2011-08-21 Published:2011-08-21

摘要: 为了进一步提高脱线中文手写体笔迹识别的正确率,提出了一种基于抗混叠轮廓波变换的特征提取算法。抗混叠轮廓波变换不仅具有轮廓波变换的多尺度、多方向特性,同时克服了轮廓波变换中频谱混叠的现象,避免了重构图像出现“划痕”现象。实验结果证明,抗混叠轮廓波变换的GGD模型与使用单小波、复小波、轮廓波变换的GGD模型方法比较,识别正确率分别提高了23.5%、7.7%、2.5%。

关键词: 小波变换, 抗混叠轮廓波变换, 广义高斯分布(GGD)模型, KL距离

Abstract: In order to enhance the precision rate of off-line Chinese handwriting-based writer identification,a new feature extraction method based on the non-aliasing Contourlet transform is presented.The transform not only has the multiscale and multidirection properties,moreover it overcomes the frequency aliasing of Contourlet transform,and avoids “scratching” phenomenon in the reconstructed image.In comparison with a single wavelet transform,the complex wavelet transform and Contourlet transform,the method increases the accuracy about 22.5%,7.7%,2.5%,respectively.

Key words: wavelet transform, non-aliasing Contourlet transform, General Gaussian Distribution(GGD) model, KL distance