中文摘要：

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%。

英文摘要：

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.