【作者】 王晓明；

【导师】 韩晓军；

【作者基本信息】 天津工业大学 ， 信号与信息处理， 2007， 硕士

【摘要】 笔迹鉴别是依据手写笔迹对书写人身份进行判别的一门技术。随着计算机技术的发展，用计算机实现笔迹鉴别可以减轻文检人员的工作压力。笔迹鉴别已成为计算机视觉和模式识别领域研究的热点问题。本文通过对改进型多通道Gabor滤波器的理论分析，最终建立起基于B／S架构的笔迹鉴别系统。首先分析了笔迹鉴别的发展历史及应用背景，并介绍了目前的技术发展状况。分析了文字提取、归一化等预处理的算法，对使用多通道Gabor滤波器实现笔迹图像的特征提取进行了理论分析。通过混合编程的方法实现，将提出的算法通过Matlab编程实现，并生成为．net组件；通过VS．net平台对Matlab中的程序进行调用，实现了两个平台之间的数据传输；在VS．net平台上实现最近邻分类器；使用ASP．net编译用户界面，并且实现网络访问，完成了笔迹图像的远程传输、笔迹特征提取模块调用、系统管理、用户登录与信息查询功能。最终整个系统实现了笔迹鉴别功能基于B／S架构的网络共享。通过实验测试，系统准确率可以达到91％，基本完成了系统设计目标。系统编程平台为MatlabR2006a和VS．net2005。更多还原

【Abstract】 The handwriting identification is used to distinguish the figure of a person. The pressure of the person who do the job of handwriting identification can be lightened by using computers to accomplish the same thing as the development of computer. The handwriting identification also becomes a hotspot problem in the domain of computer vision methodology and pattern recognition.After analyzing the theory of multi-channel Gabor wavelet, the system of handwriting identification is made based on the structure of B/S in this thesis. The history of development and the application background of handwriting identification is discussed, and the state of handwriting identification is also surveyed. About the arithmetic, the method of preprocessing are introduced, and the process of feature character distilling based on the multi-channel Gabor wavelet is also presented. The system is realized by mixed program. The arithmetic is accomplished in Matlab, and then is changed to the groupware of VS.net; the groupware is used by .net and the transport of data between the two program; the classification of nearest neighbor is made in VS.net; the UI is accomplished by using ASP.net, and at the end the function of long-distance transport, using the module of feature character comparison, system management, user identification and information demanding are accomplished. The system realizes the sharing of the function of handwriting identification based on B/S.The ratio of correctness is 91%, the goal of the system is accomplished. The whole system is accomplished in MatlabR2006a and .net2005.更多还原