Dynamic Weighting for Effective Fusion of Fingerprint and Finger Vein

Abstract: This paper investigates the feature level fusion of fingerprint and finger vein biometrics. Both features of fingerprint and finger vein are represented by minutiae which are compatible in nature. For effective fusion, redundant point elimination is implemented as the first step prior to fusion. Then the two independent feature point-sets extracted from the two biometric modalities are concatenated. For the purpose of reliable matching, the quality of features at specific interested areas is evaluated. To weaken the influence of low quality images and false features, a dynamic weighting strategy is explored based on the results of feature evaluation. Experimental results based on FVC2000 database and self-constructed finger vein database show that our scheme achieved 98.9% recognition rate, compared with single fingerprint recognition and single finger vein recognition increased by 6.6% and 9.6% respectively, compared with fusion recognition at matching score level increased by 5.4. The extensive tests on public fingerprint databases: FVC2002_DB1_A, FVC2004_DB1-A and selfconstructed finger vein database show that the dynamic weighting algorithm can achieve better performance even though poor quality fingerprint images are presented.

Keywords: Dynamic Weighting, Feature Level, Fingerprint, Finger Vein, Fusion

¹State Key Laboratory of Power Transmission Equipment & System Security and New Technology, Chongqing University, 400030 Chongqing, China yangyym@cqu.edu.cn; linkunming88@163.com, along399@126.com
²School of Computer Science and IT, RMIT University, 3001 Melbourne, Australia, fengling.han@rmit.edu.au;

[PDF]

Cite: Yongming Yang, Kunming Lin, Fengling Han, Zulong Zhang, "Dynamic Weighting for Effective Fusion of Fingerprint and Finger Vein," Progress in Intelligent Computing and Applications , vol. 1, no. 1, pp. 页码, October 2012.