基于流场与劳伦级数的指纹奇异点检测算法  

Fingerprint Singular Points Detection Algorithm Based on Flow Field and Laurent Series

作　　者：彭可[1] 刘琴[1] 刘巍 李仲阳[1] 兰浩[1] 

机构地区：[1]湖南师范大学工学院,长沙410081 [2]佛罗里达理工大学计算机科学系,美国墨尔本329016975

出　　处：《计算机工程》2013年第6期227-230,235,共5页Computer Engineering

基　　金：湖南省自然科学基金资助项目(11JJ3080);湖南省高校产业化培育基金资助项目(11CY012);湖南师范大学青年人才计划基金资助项目(ET51007)

摘　　要：提出一种基于流场与劳伦级数的指纹奇异点检测算法。运用指纹标量场的梯度信息得到指纹流场,在指纹流场与劳伦多项式之间互相关性的基础上,采用多尺度复数滤波的互相关性能量检测指纹奇异点。在SPD2010指纹数据库上进行实验研究,结果验证该算法能够实现指纹流场中奇异点的旋转不变性检测,检测效果优于SPD2010中的最优算法。A new detection algorithm of fingerprint singular points is proposed in this paper, based on flow field and Laurent series. Fingerprint orientation flow field is obtained, by using the gradient of fingerprint scalar field. Candidate singularities are detected using the cross-correlation energy calculated between fingerprint orientation flow field and complex-valued filters derived from a novel multi-scale Laurent polynomial model, which allows for rotation invariant detection of singular points in general flow fields. The algorithm is tested on SPD2010 database. Results show that the proposed algorithm outperforms the best detecting algorithm by large margin and the detection is invariant to rotational transformations.

关 键 词：指纹检测 奇异点 流场 劳伦级数 互相关性能量 旋转不变性 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]