Abstract
Owing to well-documented advantages, multimodal biometrics systems form one of the visible and promising trends in the development of biometric systems. The results produced by the multimodal biometric system come as the aggregation of results produced by individual modalities encountered in the problem. In verification systems involving a single modality, a threshold has to be identified to separate the yes-no regions of the biometrics decision. However, in multimodal verification, there are four regions separated by two thresholds. In this paper, we introduce a concept of three-valued logic in a multimodal verification system in which four thresholds and nine regions have to be identified. The thresholds (regions) are optimized for a given number of parameters. Different structures of multimodal biometrics are tested for synthetic and publicly available datasets.