计算机科学与探索 ›› 2012, Vol. 6 ›› Issue (7): 586-592.DOI: 10.3778/j.issn.1673-9418.2012.07.002

• 学术研究 • 上一篇    下一篇

构建细节点柱形结构的指纹匹配算法

付  翔+,毛紫微,刘重晋,封举富   

  1. 北京大学 信息科学技术学院 机器感知与智能教育部重点实验室,北京 100871
  • 出版日期:2012-07-01 发布日期:2012-07-02

Fingerprint Matching Algorithm by Constructing Minutia Cylinder Structure

FU Xiang+, MAO Ziwei, LIU Chongjin, FENG Jufu   

  1. Key Laboratory of Machine Perception (Ministry of Education), School of Electronics Engineering and Computer Science, Peking University, Beijing 100871, China
  • Online:2012-07-01 Published:2012-07-02

摘要: 细节点柱形编码结构是指纹匹配领域中一种新颖的三维数据结构表示。针对这种方法所面临的极大的空间存储代价和时间复杂性,以及对稀疏细节点的描述能力较弱等问题,提出了新的细节点柱形结构的匹配策略,对原有的细节点柱形编码进行改进,减少了柱形编码的计算,有效地降低了匹配时间和空间代价。同时引入了最小形变误差模型,通过对非线性形变进行度量,找回丢失的稀疏细节点对,提高了匹配的准确率。在FVC2004上的实验结果表明,新方法具有较高的效率和准确性。

关键词: 指纹匹配, 细节点柱形编码, 最小移动误差

Abstract: The minutia cylinder-code is a new representation based on 3D data structures for fingerprint matching. To solve the problems of large storage cost, time consuming and poor description for sparse minutia, this paper proposes a new evaluation strategy for minutia cylinder structures, improves the original process of minutia cylinder-code, reduces the calculation of minutia cylinder-code, and decreases the matching time consuming and storage cost. What’s more, the paper introduces the moving least squares deflation model. By measuring the deflation degree of minutia, many sparse minutia pairs can be regained. Experiments performed on FVC2004 database show both effectiveness and efficiency.

Key words: fingerprint matching, minutia cylinder-code, moving least squares