| 题名 | 大规模虹膜数据库中虹膜分类方法研究 |
| 作者 | 邱显超 |
| 学位类别 | 工学博士 |
| 答辩日期 | 2008-07-10 |
| 授予单位 | 中国科学院研究生院 |
| 授予地点 | 中国科学院自动化研究所 |
| 导师 | 谭铁牛 |
| 关键词 | 生物特征识别 虹膜分类 虹膜识别 纹理分析 Biometrics iris classification iris recognition texture analysis |
| 其他题名 | Iris Classification in Large Scale Iris Databases |
| 学位专业 | 模式识别与智能系统 |
| 中文摘要 | 随着人们对安全问题的日益重视，基于生物特征的身份识别技术逐渐受到了广泛的关注。而虹膜这种生物特征，由于其唯一性、稳定性、非接触性和高防伪特性，成为了生物特征识别领域的一个研究热点。作为一个应用性很强的研究课题，虹膜识别正从实验室走向社会应用。随着应用范围的扩大，在使用者不断增加的情况下，虹膜数据库的规模也在不断的扩大，此时进行一对多识别所需的时间也随之不断增加。虹膜分类技术将有效降低数据库搜索的时间，对提高虹膜识别系统的整体性能和实用性起着至关重要的作用。本文对虹膜分类方法进行了深入的研究，主要工作和贡献如下： 1. 为了更好的进行虹膜分类和虹膜识别研究，我们采用多种虹膜采集设备建立了CASIA V3.0虹膜图像数据库，并且将其无偿共享给国内外的科研工作者。它是目前世界上虹膜类别最多、虹膜图像数目最多的公开数据库。 2. 利用对虹膜图像的全局纹理分析来进行自动人种分类。据我们所了解，这是首次采用虹膜图像来进行人种分类的研究，本文的研究成果从模式识别和图像分析这一新颖的角度为虹膜遗传方面的生物学研究提供科学证据，从一定程度上揭示了虹膜图像和生物遗传因素之间的联系。 3. 通过实验，从统计学意义上证明虹膜图像并不是基因无关的表型特征。虹膜纹理在小尺度下，它的局部细节特征是唯一的，每一只眼睛的虹膜特征都不相同；然而在大尺度下，虹膜纹理的全局统计特征却是和基因相关的，往往相同的人种具有相似的虹膜纹理。 4. 从纹理分析的角度出发，提出了基于局部二值特征（LBP）的虹膜分类方法。 5. 提出了一种基于多通道Gabor滤波器组和机器学习算法构造虹膜纹理基元的方法。然后用虹膜纹理基元直方图来表达虹膜图像的纹理特征，并成功的将这种特征应用在虹膜分类和人种分类中。而且我们发现，不同人种中虹膜纹理基元的分布的不同才是造成亚洲人和欧洲人的虹膜图像从视觉上看起来差异很大的根本原因。 6. 在传统的虹膜识别系统中引入虹膜分类方法，大大提升了虹膜识别系统的识别速度。本文提出了一种虹膜连续分类的思想，并且巧妙的将虹膜分类算法和传统的虹膜识别算法通过构建级联的虹膜分类器整合在一起，并且在基于DSP芯片的嵌入式系统中实现。在保证虹膜识别的识别精度的情况下，大幅度提高了虹膜匹配过程中搜索数据库的速度。 |
| 英文摘要 | With the increasing requirements for security, biometrics based personal identification methods have received extensive attention. Recently, iris recognition is becoming an active topic in biometrics because of its uniqueness, stability,un-intrusiveness and high anti-forgery features. With the rapid expansion of applications, the number of users is increasing and the size of the iris database is expanding, so it needs more time for one-to-many iris recognition. Iris classification will effectively reduce the search time, that will improve the overall performance of an iris recognition system. To the best of our knowledge, this is the first attempt on iris classification. The main contributions of our work reported in this thesis are as follows: 1. In order to better carry out iris recognition and iris classification study, we employ several iris sensors to establish the CASIA V3.0 iris image database. It is now the largest open and free iris database in the world. 2. A novel ethnic classification method based on the global texture information of iris images is proposed in this thesis. Experiment results provide scientic evidence for the biological relationship between iris patterns and genetic factors. 3. We show that iris pattern is a kind of phenotypic feature with relation to the genes from statistics. At a small scale, the local features of the iris are unique to each subject, whereas at a large scale, the global features of the iris are similar for a specific race, and they seem to be dependent on the genes. 4. A novel algorithm is proposed for iris classification based on texture analysis with Local Binary Patterns(LBP). 5. Based on multi-channel Gabor filtering and machine learning, a novel method is proposed to learn a small finite vocabulary of micro-structures, which are called Iris-Textons. Then Iris-Texton histogram is used as feature vectors of iris textures, which is successfully applied to iris classification and ethnic classification. 6. By applying iris classification in iris recognition systems, the iris recognition speed is improved greatly. In this thesis, we propose a concept of continuous iris classification, and combine iris classification with traditional iris recognition algorithm by a cascade classifier. We develop this method in embedded systems, which keep high accuracy of iris identification, but reduce the search time during the matching procedure. |
| 语种 | 中文 |
| 其他标识符 | 200418014628006 |
| 内容类型 | 学位论文 |
| 源URL | [http://ir.ia.ac.cn/handle/173211/6122]  |
| 专题 | 毕业生_博士学位论文 |
| 推荐引用方式 GB/T 7714 | 邱显超. 大规模虹膜数据库中虹膜分类方法研究[D]. 中国科学院自动化研究所. 中国科学院研究生院. 2008. |
| 个性服务 |
| 查看访问统计 |
| 相关权益政策 |
| 暂无数据 |
| 收藏/分享 |
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。
修改评论