Fusion of Auxiliary Information for Multi-modal Biometrics Authentication

In this paper, we propose to use the multivariate polynomials for fusion of fingerprint and voice verification systems. On top of using just the fingerprint and voice data for fusion, additional information such as fingerprint image quality and deliberate control inputs are also integrated within the same framework. Numerical experiments using three multivariate polynomial models show significant improvements of matching performance as compared to the optimal weighting method in terms of receiver operating characteristic curves.