脱机汉字签名鉴别的多特征多分类器设计

针对在脱机条件下的手写汉字签名计算机鉴别问题,提出了一种基于多类特征组合分类器综合鉴别的方法,该方法对签名图像提取了3种类型的特征向量,分别输入到BP神经网络加以分类,组合各分类结果得到最终输出值。实验结果表明,与采用单一类型特征或者采用单一分类器相比,该系统具备较高的鉴别率,效果令人满意。

Off-line Chinese Handwriting Signature Verification Based on Different Kinds of Features and Classifiers

For the problem of off-line Chinese handwriting signature verification,this paper presents a methodology of integrated identification based on different kinds of features and classifiers.In this system,three kinds of features are extracted from the signature images and classified by their single BP ANN(Artificial neural network).Compared with single feature or classifier,the proposed algorithms in our system are proved to be more effective.