首页 | 本学科首页   官方微博 | 高级检索  
     

一种有效的基于证据理论的离线签名识别方法
引用本文:陈刚,李弼程,曹闻,刘安斐. 一种有效的基于证据理论的离线签名识别方法[J]. 计算机工程与设计, 2006, 27(17): 3256-3257,3260
作者姓名:陈刚  李弼程  曹闻  刘安斐
作者单位:信息工程大学,信息科学系,河南,郑州,450002;信息工程大学,测绘学院,河南,郑州,450002
摘    要:提出了一种有效的基于证据理论的离线签名识别方法。从签名图像的3种信息载体中提取出4种特征,利用所提取的4种特征分别构造基于证据理论的k-NN分类器对签名图像进行初步识别,将各k-NN分类器的输出作为证据,用改进的证据理论合成公式融合不同分类器的输出得到最终识别结果。结果表明:该识别方法能有效地提高离线签名的识别率。

关 键 词:离线签名识别  证据理论  预处理  特征提取  k-NN分类器  融合
文章编号:1000-7024(2006)17-3256-02
收稿时间:2005-07-03
修稿时间:2005-07-03

Effective off-line signature recognition method based on evidence theory
CHEN Gang,LI Bi-cheng,CAO Wen,LIU An-fei. Effective off-line signature recognition method based on evidence theory[J]. Computer Engineering and Design, 2006, 27(17): 3256-3257,3260
Authors:CHEN Gang  LI Bi-cheng  CAO Wen  LIU An-fei
Affiliation:1. Department Of Information Science, University of Information Engineering, Zhengzhou 450002, China; 2. Institute of Surveying and Mapping, University of Information Engineering, Zhengzhou 450002, China
Abstract:An effective method based on evidence theory for off-line signature recognition is brought forward. Firstly, four sets of features of a signature are extracted from its three feature images, and then these features are fed to four k-NN classifiers for elementary recognition. Finally, recognition result is obtained by fusing the previous recognition results from different classifiers with a modified combination rule of evidence theory. Experimental results show that the proposed method is effective.
Keywords:off-line signature recognition   evidence theory   preprocess   feature extraction   k-NN classifier   fusion
本文献已被 CNKI 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号