CASIA OpenIR  > 毕业生  > 硕士学位论文
基于关联概率模型的指纹识别算法
Alternative TitleFingerprint Identification Based on Correlative probability Model
何余良
Subtype工学硕士
Thesis Advisor田捷
2003-05-01
Degree Grantor中国科学院研究生院
Place of Conferral中国科学院自动化研究所
Degree Discipline模式识别与智能系统
Keyword生物识别技术 关联概率模型 关联关系 可变界限盒 综合细节点 Biometrical Identification Technology Correlative Probability Model Relative Relationship Invariable-bounding Comprehensive Minu
Abstract近十几年的生物识别技术的迅猛的发展和广泛的应用前景已经吸引了许多 研究者的极大的兴趣。而基于指纹身份识别技术也因其良好的可靠性而逐渐应 用到人们生活中。科研工作者对于基于指纹生物识别技术的研究方兴未艾,有 关指纹识别的产品和新技术也日新月异。 指纹识别是寻找一个最佳的形变模型来模拟指纹图像中形变(非线性或线 性),使得两幅指纹图像之间的差异最小。因此,如何提取指纹图像的特征信息、 建立最优的形变函数、以及定义指纹特征信息之间的差异评估函数一直是影响 识别算法性能的重要因素,也是从事指纹识别研究工作者所需解决的难点。虽 然,人们已经提出很多种方法来解决这些问题,但都有其局限性。因此,本文 借助大数定律的思想,来分析和设计本文的自动指纹识别算法,并提出了一种 基于关联概率模型的指纹识别算法。在本文中与其它算法的不同之处在于: 第一,以多尺度原理对指纹图像进行增强处理。 第二,根据指纹特征和纹理模式的独特性,定义了综合细节点和细节点之问 的关联关系等来描述指纹图像的唯一性。 第三,利用指纹的综合信息,构造指纹的局部拓扑结构,建立了一个相似度 与形变的关联概率模型来动态分析指纹图像的形变。 第四,对影响算法性能的各种参数进行统计分析,采用基于最大熵原则的多 闽值方法搜索出最优的性能参数。 基于FVC数据库的实验结果表明本文提出的基于关联概率模型的指纹识别 算法是有效性。
Other AbstractThe biometric identification technology has attracted interests of researchers for its rapid development. And the reliability of fingerprint-based identification techniques has made it to be widely used in civilian application. The matching between two fingerprints is to find an optimal deformation transformation model to simulate the deformations in fingerprints and minimize the difference between the two images. Therefore, extracting fingerprint feature to represent a fingerprint, building an optimal deformation model, and defining a measurement function of fingerprint features are very important parts of a fingerprint identification method, in which the researchers have been involved. Although many methods have been proposed to solve these problems, most of them are only designed to solve some special problems and not analyze and solve all mentioned problem about a fingerprint identification method. We statistically design our fingerprint method based on correlative probability model. The novel ideas of this method are listed as follows: First, we define comprehensive minutiae, including minutiae and their associated texture, to present a fingerprint. Second, we construct a local feature structure with the comprehensive minutiae to build a similarity-deformation probability model to dynamically analyze deformations in a fingerprint. Third, we use the multi-level threshold selection by maximum entropy criterion to set the thresholds used in our method. Finally, a dynamic scale apace method is used to enhance fingerprints. Compared with other methods in literature and our previous invariable-bounding method, our method is more efficient and effective.
shelfnumXWLW697
Other Identifier697
Language中文
Document Type学位论文
Identifierhttp://ir.ia.ac.cn/handle/173211/6789
Collection毕业生_硕士学位论文
Recommended Citation
GB/T 7714
何余良. 基于关联概率模型的指纹识别算法[D]. 中国科学院自动化研究所. 中国科学院研究生院,2003.
Files in This Item:
There are no files associated with this item.
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[何余良]'s Articles
Baidu academic
Similar articles in Baidu academic
[何余良]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[何余良]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.