银川市专业模仿笔迹签名

近年来，生物识别技术取得了飞速的发展，手写体笔迹识别技术是其中的一个非   常活跃的研究领域。作为一种身份鉴别的手段，笔迹是一种稳定的行为特征，每一个   人的笔迹都具有一定的不变性，而且笔迹的获取具有非侵犯性，易为人所接受。

正是由于笔迹的这些特点，使它广泛应用于刑事侦察，计算机登录，信用卡签字和电子商   务等领域。目前，这个领域的研究重点主要集中于签名验证，而签名验证的算法要求   识别文字和训练文字具有相同的内容，这样就存在模仿和伪造的可能性。本文通过在   已有笔迹鉴别技术的分析之上，提出了几种有效的一般意义下的书写人笔迹鉴别方法。

本论文的主要内容如下：       

第一章首先介绍了基于笔迹的身份鉴别技术的背景和有关概念，着重介绍了当  前有代表性的笔迹鉴别方法。       

第二章主要介绍在线笔迹鉴别的数据采集方式，并在借鉴已有字符识别预处理   技术的基础上，针对笔迹数据预处理的一些常用的算法进行了阐述与分析，其中主要   包括剔除飞点和虚假抬笔点，进行字符图象的人小归一化，以及笔迹图象的倾斜校正。       

第三章实现了基于纹理分析的与内容无关的笔迹鉴别方法，并给出了几种纹理  分析方法用于笔迹鉴别的比较。       

第四章提出了一种新的基于特征字融合的笔迹鉴别方法，通过用PCA方法在灰   度笔迹图像上进行学习，可以找出代表不同人书写习惯的特征字，而特征字在PCA  空间上的一组带有加权参数的投影矩阵则代表了这幅笔迹的整体书写特征。      

第五章给出了一种文本无关的在线笔迹鉴别方法，利用笔迹图象的形状特征和  动态特征对笔迹进行鉴别。       

第六章对本论文的工作进行了总结并给出了笔迹鉴别的展望。

Biometrics has been an active research area for a long time aiming at automatic  identity recognition based on individual physiological or behavioral characteristics.  Personal identification based on handwriting is a kind of behavioral biometric  identification approach. Each person has his individual writing style and the handwritings  are easy to obtain. Handwriting based personal identification has a wide variety of  potential applications, from security, forensics, financial activities to archeology. For this  reason, much research has touched on this field, but most of them rested on signature  verification, which has the disadvantage in that the identification content is fixed and   limited, making it prone to forgery. This master thesis proposes several efficient methods   for writer identification through deep analysis of the problem nature. The thesis is  organized as follows.        

Chapter 1 introduces some concepts about handwriting identification (HI) and the significance, application background, difficulties and evaluation criteria of HI technology.  Also the most popular methods of HI are presented in this chapter.        

Chapter 2 describes the data collection manner and mainly focuses on the  preprocessing processes of the on-line and off-line handwriting data, which include  infliction point and false pen-lift removal, character normalization and handwriting image  slant correction method.        

Chapter 3 realizes the text-independent HI based on texture analysis, and compare the   performance of them.       

 A novel algorithm is presented for writer identification in chapter 4. Principal  . Component Analysis is applied to the gray-scale handwriting images to find a set of   individual words which best characterize a person's handwriting style and have maximal   difference from other people style. During identification, we only need to utilize a set of   individual characteristic words for comparison, instead of comparing the whole   handwriting text to identify the writers.         

Chapter 5 gives another text-independent on-line HI method, which mainly utilizes   the visual shape features of character and dynamic features for handwriting identification.        

Chapter 6 summary the whole paper and give a prospect of the development of HI.