Score-Level Fusion in Biometric Verification

This paper presents and details the investigations into the effectiveness of various score level fusion approaches in both multi-sample and multimodal biometric systems. Experimental investigations are conducted using face and speech modalities. The scores for face and voice biometrics are based on the use of different features extracted from the XM2VTS database. These scores are provided by idiap research institute. Experimental results demonstrated that a higher performance can be achieved by applying the fusion at multimodal level.