CASIA OpenIR  > 毕业生  > 硕士学位论文
Alternative TitleFingerprint Image Matching
Thesis Advisor蒋田仔
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 指纹 方向场 细节特征 配准 匹配 Biometrics Fingerprints Orientation Field Minutia Registration Matching
Abstract随着社会信息化的日益发展,身份鉴别方法已经日益显示出其重要性与必要性。由于 传统的身份鉴别方法已经无法跟上时代发展的需要,因此人们将目光更多的投向了基于生 物特征的身份识别,这其中最成熟、可靠、应用最广泛的是指纹识别。 本文作者在模式识别国家重点实验室攻读硕士学位期间,对指纹匹配算法的进行了深 入的研究。在本文中.主要的工作和贡献有: ①提出了指纹的方向场作为一个稳定的指纹全局特征可以应用到指纹匹配中,使匹配 算法将指纹的全局特征和局部特征很好的结合起来。 ②提出了一种面向于IC卡的指纹匹配算法。该算法充分考虑了IC应用系统在运算速 度及存储容量上的限制,把指纹的方向场作为新的特征加入到配准和计算匹配值 中,使算法结合了局部特征(特征点)和全局特征(方向场)。该算法在一定程度 上解决了当特征点较少时,只采用局部特征无法分辨的问题,在只增加少量的存储 量和计算量的情况下提高了匹配的准确性。 ③提出了一种基于互信息的指纹匹配算法。我们首先从模板和输入指纹的方向场提取 出方向特征.然后通过最大化它们之间的互信息的方法来实现指纹的精确配准,最 后通过比较特征点和互信息给出匹配结果。我们采用了一种由粗到精,多级化的方 法来加快指纹配准的速度。算法的最大优点是它与人们比较指纹的过程相似:首先 看整体纹路是否一致,然后再比较局部的细节特征是否相同。 本论文首先介绍了自动指纹身份识别系统的一些基本概念、系统组成和系统性能的评 价方法。然后,对指纹匹配算法给出了一个较为全面的综述。接着,详细介绍了我们所提 出的面向于IC的指纹匹配算法和基于互信息的指纹匹配算法。最后,对我们的工作进行 了总结并对今后的工作作出了一些展望。
Other AbstractAccurate personal identification is becoming more and more important to the operation of our ever increasingly electronically inter-connected information soci- ety. Traditional personal identification technologies cannot satisfy the increasing security requirements. Due to its uniqueness and stability, biometrics has become the most promising way to meet such requirements. Being the oldest and the most mature biometrics, fingerprint is the most widely used personal identifica- tion method. In this thesis, we study the fingerprint image matching algorithms. The main contributions of this thesis include the following issues: ① The orientation field, which is a stable global features of fingerprint images, is employed in fingerprint matching, making matching algorithms integrate global features with local ones. ② An IC card oriented fingerprint matching algorithm is presented. Taking full consideration of the limitations of processing speed and storage space of IC card systems, this algorithm makes use of orientation field of fingerprint images as a new feature in the registration and computation of matching scores. Combining local features (minutiae) with global ones (orientation field), this algorithm tries to solve the problem that fingerprints cannot be discriminated solely using minutiae when relatively few minutiae available, making it possible to increase matching accuracy at the cost of little increase of storage and computation cost. ③ A mutual information based fingerprint matching algorithm is presented. This algorithm finds the accurate alignment by maximization of mutual in- formation between features extracted from orientation field of template and input fingerprint images. Then matching result is jointly given by minu- tiae matching and maximal mutual information achieved in registration. To speed up registration, we takes a hierarchical method. The prominent char- acteristics of this algorithm is that it much resembles human behavior of comparing fingerprints, which results in reduced occurrences of misalign- ment and improved registration accuracy. In this thesis, we first give an introduction to the fundamental definitions, system architecture and evaluations of automatic fingerprint identification sys- tems (AFIS). Then, a comprehensive survey is presented on the state of the art of fingerprint image matching algorithms.Afterwards,we detailed our IC card oriented fingerprint matching algorithm and mutual information based fingerprint matching algorithm.Finally,a summary of our work and a vista of future research are presented.
Other Identifier683
Document Type学位论文
Recommended Citation
GB/T 7714
刘力锋. 指纹图像匹配算法的研究[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:
File Name/Size DocType Version Access License
硕士生学位论文-683.pdf(9190KB) 暂不开放CC BY-NC-SAApplication Full Text
Related Services
Recommend this item
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[刘力锋]'s Articles
Baidu academic
Similar articles in Baidu academic
[刘力锋]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[刘力锋]'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.