Fusing multiple matcher's outputs for secure human identification

Multimodal biometrics is an emerging area of research that aims at increasing the reliability of biometric systems through utilising more than one biometric in decision-making process. An effective fusion scheme plays a key role in combining the information presented by the multiple domain experts. Such information can be integrated at several distinct levels, such as sensor level, feature level, match score level, rank level and decision level. This paper describes the combination process of different monomodal expert through rank and decision fusion methods using iris, ear and face biometrics for secure human authentication. For rank-level fusion, plurality voting, Borda count and logistic regression approaches are employed and compared, and for decision-level fusion, AND/OR, majority voting, weighted majority voting and behavioural knowledge space approaches have been implemented and tested. The key contribution of the paper is in comparison of the recognition performance of the developed multimodal system for all of the above approaches. The results indicate that fusing individual modalities improve the overall performance of the biometric system and the logistic regression rank-level fusion results in the highest recognition performance even in the presence of low-quality data.