Biometric fusion for palm-vein-based recognition systems

Abstract

In this paper we investigate the impact on performance of biometric fusion techniques for palm-vein-based recognition systems. In more detail, feature-level fusion, score-level fusion as well as decision-level fusion approaches are applied in a biometric system exploiting patterns of palm veins for user recognition, and both local binary pattern and local derivative pattern features for template generation. The obtained results show that a significant performance improvement can be achieved when the aforementioned feature extraction approaches are jointly taken into account, compared to the case where no biometric fusion is performed.