基于FTP的3D掌纹识别技术研究-计算机科学与技术专业论文.docxVIP

Classified Index: TP391.43 UDC: 004.93 Thesis for the Master Degree of Engineering 3D Palmprint Identification Research Based On FTP Candidate: Shouyu Ma Supervisor: Prof. Zhang Dapeng Academic Degree Applied for: Master of Engineering Specialty: Computer Science and Technology Affiliation: Shenzhen Graduate School Date of Defence: December, 2011 Degree-Conferring-Institution: Harbin Institute of Technology 摘要 传统的掌纹识别主要是基于掌纹二维图像，从图像中提取掌纹纹理信息作为 特征来进行身份识别，然而二维掌纹识别存在以下缺陷：受光照影响严重；手掌 颜色变化或污渍会影响识别精度；存在被复制的风险。随着三维测量技术的不断 发展，使用三维掌纹信息进行识别成为可能，三维掌纹除了有良好的防伪能力外， 在对光照、手掌颜色变化等方面，也有着先天的优势，三维掌纹识别将成为一个 重要的研究领域。 本文对三维掌纹重构和识别进行了系统的研究，本文的主要研究内容包括： 三维掌纹采集系统的设计与搭建。常用的三维成像技术有立体视觉法、激光 测量法、结构光法等，综合考虑设备成本、采集速度等因素，我们采用傅里叶变 换轮廓术来重构三维掌纹，根据其原理设计搭建了三维掌纹采集系统，建立了包 括 100 个手掌、1200 幅掌纹图像的数据库。 基于 FTP 的三维掌纹重构。使用傅里叶变换轮廓术重构三维掌纹需要经历以 下几步：图像预处理、傅里叶变换、频域滤波、逆傅里叶变换、计算相位、相位 展开和计算高度。最后将采集的样本生成三维掌纹样本库。 掌纹深度特征提取及分类。三维掌纹的深度特征不同于二维掌纹的纹理特征， 是三维掌纹特有的，本文主要定义了以下四类深度特征：掌纹最大深度、掌纹分 区域深度均值、等位横截面面积、射线长度。这些特征提取之后，整体特征的维 数较高，为了提高识别效率，用正交线性判断进行降维，最后使用降维后的特征 进行分类，取得了良好的效果。 关键词：生物特征识别；傅里叶变换轮廓术；三维掌纹识别；结构光三维成像 I Abstract 2D palmprint identification is based on 2D image and extract texture information as features for identification. However, 2D palmprint identification has some shortcoming: Firstly, light seriously affect recognition accuracy. Secondly, the change of palm color or stain will affect recognition accuracy. Thirdly, there is the risk of being copied. With the continuous development of three-dimensional measurement techniques, it is possible to use 3D palmprint information for identification. 3D palmprint has a good-security capabilities. It also has inherent advantages in the light changes and the color changes of hands. Three-dimensional palmprint identification has become an important research area. In this paper, we have a depth study on the reconstruction of three-dimensional palm. Then we develop and build 3D palmprint colleting system. At last, we extract 3D palm feature and do identification. The research include