计算机工程与应用 ›› 2009, Vol. 45 ›› Issue (36): 161-164.DOI: 10.3778/j.issn.1002-8331.2009.36.048

• 图形、图像、模式识别 • 上一篇    下一篇

用于手写签名识别的小波包混合高斯模型

肖春景1,李春利1,2,乔永卫3,张 敏1   

  1. 1.中国民航大学 计算机科学与技术学院,天津 300300
    2.哈尔滨工程大学 自动化学院,哈尔滨 150001
    3.中国民航大学 工程技术训练中心,天津 300300
  • 收稿日期:2008-12-25 修回日期:2009-03-02 出版日期:2009-12-21 发布日期:2009-12-21
  • 通讯作者: 肖春景

Wavelet packs and Guass model for off-line handwritten signature recognition

XIAO Chun-jing1,LI Chun-li1,2,QIAO Yong-wei3,ZHANG Min1   

  1. 1.Computer Science and Technology Department,Civil Aviation University of China,Tianjin 300300,China
    2.Automation Department,Harbin Engineering University,Harbin 150001,China
    3.Engineering and Technical Training Center,Civil Aviation University of China,Tianjin 300300,China
  • Received:2008-12-25 Revised:2009-03-02 Online:2009-12-21 Published:2009-12-21
  • Contact: XIAO Chun-jing

摘要: 为了解决手写体签名识别中数据预处理复杂、分割困难、特征提取不充分等问题,提出了基于小波包分解与高斯模型的脱机手写体签名识别方法。它对归一化的整个签名图像进行小波包分解,对分解值自动聚类和高斯建模,不但没有去噪、旋转、平移、分割的过程,而且特征提取完全且是可逆的分解过程。实验表明提出方法比其他方法具有更好的抗噪性、鲁棒性、适应性和识别率,为含噪脱机手写体签名识别提供了一种可行的技术解决方案。

关键词: 小波包, 高斯模型, 标准化, 聚类

Abstract: The paper proposes a way of off-line handwritten signature recognition based on wavelet packs and Gauss model in order to solve the problems of complicated data preprocessing,difficult segmentation and deficient feature extraction.It decomposes the whole signature image with wavelet packs after normalized,clusters the decomposed values and gains Gauss modeling.It doesn’t need noise reduction,rotation,translation and segmentation,and feature extraction is complete and is reversible decomposed course.The results show that the method proposed has better antinoise,flexibility than other methods and improves the recognition rate.It provides a feasibility techno-project for off-line handwritten signature.

Key words: wavelet packs, Guass model, normalization, clustering

中图分类号: