Journal of Frontiers of Computer Science and Technology ›› 2017, Vol. 11 ›› Issue (4): 643-651.DOI: 10.3778/j.issn.1673-9418.1602033

Previous Articles     Next Articles

Fingerprint Registration Based on Finite Element

YAO Lisha+, CHENG Jiaxing   

  • Online:2017-04-12 Published:2017-04-12



  1. 安徽新华学院 信息系统软件研究所,合肥 230088

Abstract: Fingerprint registration algorithm is studied for the complexity and variety of fingerprint deformation. Because the existing fingerprint registration algorithms have the defects of high complexity, low precision, slow speed and easy to be affected by fingerprint deformation, this paper introduces finite element analysis theory, and proposes a fingerprint registration algorithm based on finite element. In this algorithm, discrete finite element is used as basic unit to simulate the deformation of whole elastomer for complex deformation in fingerprint images. It takes the turning point, bifurcation point, the number of ridge lines between two points in fingerprint images and so on as the features. The fuzzy similarity is creatively proposed as metric for solving the overall stiffness equation to complete fingerprint registration. The experimental results show that the error of displacement of the proposed algorithm is the smallest and the time is 8.894 seconds, it lowers the complexity, and also improves the precision. At the same time, it can avoid the influence of the complexity and variety of fingerprint deformation on the accuracy of the fingerprint registration algorithm.

Key words: fingerprint registration, finite element, fuzzy similarity, fingerprint feature points

摘要: 针对指纹图像中形变复杂多样的特点,对指纹图像配准算法进行研究,发现目前指纹配准算法存在复杂度高,精度低,速度慢,易受指纹形变等影响的缺陷,引入有限元分析理论,提出了有限元指纹图像配准算法。该算法将指纹图像配准中的复杂形变转换为有限个离散形变单元组成的弹性形变,以分叉点、转折点、指纹图像上两点的连线所穿越的脊线数量等特征作为特征指标,求出采样指纹与标准指纹的模糊贴近度,并创造性地以此为相似性测度,构造总体刚度方程,进而迭代求解最佳位移,完成指纹图像配准。实验表明,该算法位移量的误差最小,用时为8.894 s,其有效降低了算法复杂度,提高了精度,同时也有效避免了指纹复杂多样的形变等因素对指纹配准算法精度的影响。

关键词: 指纹配准, 有限元, 模糊贴近度, 指纹特征点