用于手写签名识别的演化超网络

手写签名作为易被大众所接受的生物特征身份认证方式,已成为模式识别领域一个重要研究热点.针对现有手写签名存在易模仿难鉴定的问题,提出一种结合演化超网络模型的手写签名认证方法.为了平滑噪声,构造出可读性强的笔迹特征集,采用向量化和平滑采集点的方法对手写签名样本进行预处理,从而提取出位置和方向特征属性,采用演化超网络模型对签名进行学习和鉴定.为验证该方法的有效性,对20个签名用户分别采集了40个真实签名和20个伪造签名数据进行实验.实验结果表明,该方法对用户签名的误拒率(false rejection rate,FRR)为4.75％,误纳率(false acceptance rate,FAR)为3.75％,识别率(verification accuracy,VA)为95.75％.同时和其他传统的识别算法相比,具有更高的识别率.

Evolutionary hypernetwork for handwritten signature verification

Handwritten signature is one of the most widely accepted biometric verification technologies and it has become an important research focus in the field of pattern recognition. Handwritten signature is a challenging task because of the easi-ness of forging and difficulty of identifying one' s signature. As for these problems, an evolutionary hypernetwork model for handwritten verification is proposed in this paper. In order to smooth noise and construct the more readable handwritten fea-tures, the position of points and direction are extracted as features through the vectoring and smoothing as pre-processing. Then, an evolutionary hypernetwork is utilized for handwritten signature learning and verification. Finally, in order to test the efficiency of the proposed method, a signature dataset with 40 genuine signatures and 20 forged signatures of 20 writers is employed. The experimental results show that the false rejection rate ( FRR) , the false acceptance rate ( FAR) , and the verification accuracy ( VA) of the proposed method are 4.75％, 3.75％, and 95.75％, respectively. Moreover, the proposed model achieves higher classification accuracy than other traditional pattern recognition method.