基于集成神经网络的离线手写签名鉴别方法

离线手写签名鉴别是一种基于生物特征的身份识别技术，利用集成神经网络对手写签名进行自动鉴别。该集成神经网络由特征分配网络、神经网络认证主体和决策融合识别网络三部分构成。通过为每个签名者构造独立的分类器，并分别为每个分类器构造认证、识别训练集，解决了离线签名的认证和识别问题。基于此方法的签名鉴别实验获得了比较满意的结果，可以用来辅助人类专家进行签名鉴别。

Off-line Chinese signature authentication based on integrated neural networks

To assist the signature specialists, the authentication system for off-line Chinese signatures was developed based on a new Integrated Neural Network (INN), which was composed of the modules, namely Feature Distributed Network (FDN), Neutral Networks Classifier (NNC) and Decision Fusional Network (DFN). In the method, a specific classifier was created for each of the signatures,and both training sets were constructed to solve the verification and recognition problems. The proposed method yields satisfactory...