基于特征选择的脱机签名模糊鉴定建模与仿真  

Off-line Signature Fuzzy Verification Modeling and Simulation Based on Feature Selection

作　　者：田伟[1] 乔谊正[1] 马志强[1] 

机构地区：[1]山东大学控制科学与工程学院,济南250061

出　　处：《系统仿真学报》2008年第7期1734-1739,共6页Journal of System Simulation

基　　金：国家自然科学基金(60775023);山东省自然科学基金(Y2002G02;Z2005G03)

摘　　要：在建立脱机签名鉴定模型的过程中,首先设计提取了签名的静态特征和伪动态特征,并提出了一种基于离差测度的新的选择准则用于特征选择。另外,在鉴定模型中引入模糊理论,通过模糊集合来表征特征的不确定性,并采用隶属度函数构造权重系数作用于模糊规则的输出来进行鉴定。仿真实验对于中、英文两种签名数据库分别取得了大约9.2%和10.5%的平均错误率,其中,英文数据库的平均错误率较现有文献的结果降低了大约3%。During building the model of off-line signature verification, static and pseudodynamic features are extracted firstly Then a new criterion based on discriminative measure is proposed for feature selection. In addition, the model introduces fuzzy theory to verification where the variation of selected features is described by fuzz）＇ sets and the weight coefficient is devised by the membership functions to reflect the contribution to verification outputs. Two databases of Chinese and English signatures are applied to the experiments and the average error rates of 9.2% and 10.5% are obtained approximately, and about 3% for English database is reduced when compared with the other existing results.

关 键 词：脱机签名鉴定 特征提取 特征选择 模糊模型 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]