【作者】 李彬；

【导师】 张大鹏； 王宽全；

【作者基本信息】 哈尔滨工业大学 ， 计算机应用技术， 2006， 博士

【摘要】 信息社会对系统和信息安全性的要求日益增加,需要对人的身份进行鉴别的应用场合越来越多,传统的身份鉴别方法由于其自身所固有的弱点已不能够满足社会发展的要求。在这种情况下生物识别技术应运而生。生物识别技术是利用人体所固有的生物特征来进行自动身份识别或鉴别。常见的人体生物特征包括:指纹、人脸、虹膜、掌纹、语音、签名等,这些人体生物特征通常具有“人人拥有、人人不同、长期不变”的特点,并且不易遗忘和丢失,也难以伪造和模仿,所以很适合用来进行身份识别或鉴别。联机手写签名鉴别是生物识别技术一个重要分支。签名鉴别具有其独特的优点:手写签名自古以来就是一种被人们普遍认可并广泛应用的行为特征;手写签名的采集设备价格比虹膜和掌纹等采集设备更低廉;作为一种行为特征,手写签名比人体物理特征更难于模仿等。因此,联机手写签名技术一直是生物识别技术领域的研究热点。本文对联机手写签名鉴别技术进行研究,主要研究内容包括:1.基于改进动态时间规整和一维曲线段弹性匹配的联机手写签名鉴别。联机手写签名可以看成是一个等时间间隔的序列,通过一些简单的计算,可以得到多条一维曲线来代表原始签名。对于一维曲线,在分段和特征描述方面更加简单,而且通常手写签名在x和y方向的稳定性是不相同的,通过一维的描述可以很容易地将签名分解为不同稳定性的特征曲线,并在签名鉴别中赋予不同的权重。因为曲线段的特征既包括端点的特征又包括曲线的特征,采用传统的动态时间规整无法解决因误分段所带来的匹配误差,因此,本文提出了一种基于后向合并的改进动态时间规整算法,该方法较好地解决了误分段问题。2.对二维签名笔段特征空间的稳定性进行分析。签名鉴别是一个无真实伪造样本的特殊的两类模式分类问题,因此对真实签名样本空间进行稳定性分析就显得非常有意义。考虑到签名笔段的实际意义,本文将签名在二维空间进行分段,然后提出一种稳定段提取的算法来构造稳定段特征矩阵。并提出了一种与通用的主分量分析(Principal Component Analysis, PCA)截然相反的方法——基于零分量分析(Null Component Analysis, NCA)和主分量分析的签名特征空间稳定性分析方法。更多还原

【Abstract】 System and information security is becoming increasingly important in the information society. Personal authentication is becoming necessary in more and more fields. The traditional personal authentication methods cannot keep up with the development of the society because of their inherent defects. Under such circumstance, biometrics emerged as the time requires.Biometrics are the technologies that analyze human characteristics for automated personal authentication. Many common biometric features, such as fingerprint, face, iris, palmprint, voice, signature etc., have the properties that they are possessed by everyone, are different with different person and keep stable in a long period. They cannot be stolen nor lost, fake and are difficult to be forged or imitated. Therefore, they are very suitable for the personal authentication.Online handwritten signature verification is an important branch of biometrics. Handwritten signature has its own virtues: handwritten signature has been a human behavior characteristic and been widely accepted and applied since ancient times; online signature capture devices are much cheaper than iris and palmprint devices; handwritten signature is more difficult imitated than other personal physical characteristics. Therefore, online handwritten signature verification is hotspot in the biometrics field.The main purpose of this dissertation is to research the technology of parameter-based online signature verification. The research includes following parts:1. Online signature verification based on improved dynamic time warping and the elastic matching of 1D curve segments. An online signature can be considered as a point sequence with a same interval, and can be described by several 1D curves after some simple transformations. The segmentation and feature description of a 1D curve is very simple. It is very easy to separate a signature into several curves with different stableness by this 1D description. In the verification, different curves are given different weights. Because a 1D segment contains not only the features of the endpoints but the inner feature of更多还原