Score-Based Fusion in Multi-Unit Biometric Recognition System

Abstract:

The performance of biometric systems can be improved by combining multiple units through score level fusion. In this paper, different fusion rules based on match scores are comparatively studied for multi-unit fingerprint recognition. A novel fusion model for multi-unit system is presented first. Based on this model, we analyze the five common score fusion rules: sum, max, min, median and product. Further, we propose a new method: square. Note that the performance of these strategies can complement each other, we introduce the mixed rule: square-sum. We prove that square-sum rule outperforms square and sum rules. The experimental results show good performance of the proposed methods.

• Periodical: Applied Mechanics and Materials (Volumes 48-49)
• Pages: 1010-1013
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.48-49.1010
• Online since: February 2011
• Authors: Yong Li, Jian Ping Yin, En Zhu
• Keywords: Biometric Recognition, Multi-Unit, Score-Based Fusion, Square Rule

References

[1] Ross A, Nandakumar K, and Jain A K, in: Handbook of Multibiometrics [M]. New York: Springer Verlag, (2006).

[2] Rowe R K, Uludag U, Demirkus M, Parthasaradhi S, Jain A K, in: A Multispectral Whole-Hand Biometric Authentication System, Biometrics Symposium, 2007: 1-6.

DOI: 10.1109/bcc.2007.4430532

[3] Kumar and Zhang D, in: Combining fingerprint, palmprint and hand-shape for user authentication, Proc. Int'l. Conf. Pattern Recognition (ICPR), 2006: 549-552.

DOI: 10.1109/icpr.2006.383

[4] Kittler J, Hatef M, Duin R P W, and Matas J, in: On combining classifiers. IEEE Trans. on Pattern Anal. Machine Intell., 1998, 20(3): 226–239.

DOI: 10.1109/34.667881

[5] Alkoot F M, Kittler J, in: Improving the performance of the product fusion strategy, Proc. Int'l. Conf. Pattern Recognition (ICPR), 2000, (2): 549-552.

DOI: 10.1109/icpr.2000.906040

[6] Tax D, Breukelen M. van , Duin R, in: Combining multiple classifiers by averaging or by multiplying, Pattern Recognition 2000, (33): 1475–1485.

DOI: 10.1016/s0031-3203(99)00138-7

[7] Prabhakar S, Jain AK, in: Decision-level fusion in fingerprint verification, Pattern Recognition, 2002, (35): 861-874.

DOI: 10.1016/s0031-3203(01)00103-0

[8] Garris M D, Watson C I, and Wilson C L, in: Matching Performance for the US-Visit IDENT System Using Flat Fingerprints. Technical Report 7110, National Institute of Standards and Technology (NIST), July (2004).

DOI: 10.6028/nist.ir.7110

[9] Sedgwick N, in: Preliminary Report on Development and Evaluation of Multi-Biometric Fusion using the NIST BSSR1 517-Subject Dataset, May (2005).

[10] Ulery B, Hicklin A, Watson C, Fellner W, Hallinan P, in: Studies of Biometric Fusion, NISTIR 7346, 9. (2006).

DOI: 10.6028/nist.ir.7346

[11] Maltoni D, Maio D, Jain A K, Prabhakar S, in: Handbook of Fingerprint Recognition, Springer Verlag, (2006).

[12] http: /bias. csr. unibo. it/fvc2006.

[13] Lee K, Park KR, Jang J, Lee S, and Kim J, in: A Study on Multi-unit Fingerprint Verification, AVBPA 2005, LNCS 3546, 2005: 141–150.

DOI: 10.1007/11527923_15