期刊名称：International Journal of Advanced Research in Computer Engineering & Technology (IJARCET)
印刷版ISSN：2278-1323
出版年度：2013
卷号：2
期号：9
页码：2586-2594
出版社：Shri Pannalal Research Institute of Technolgy
摘要：Signature verification and recognition system can be used as a security system such as verification for assessing entry application, cheques, bank transactions and password substitutions. In this paper a method of off-line (static) signature verification has been developed where signature samples of each person was cropped by automatic cropping system and here proposed methods of feature extraction that is projection based local features. The approach of signature verification based on Discrete Radon Transform (DRT) by using only three genuine signatures samples of each person has been employed here in order to train our system. In the proposed system process time required for training our system for each person is between 0.6 to 2.8 seconds and requires less memory for storage. The experimentation of the proposed system was performed on a large signature database (genuine and forgeries) for over 150 signature images. The proposed technique gives FAR of 6.66%, FRR of 4.44% and EER 5.19%.
关键词：Off-line Signature verification; on-line signature verification; ; biometrics; authentication systems ; ; Median filter; Local features; ; Discrete Radon transform
