Biometric data fusion based on subjective logic

Biometric identification and authentication is affected by varying levels of reliability depending on the inherent strength of a specific biometric modality, the specific techniques used to analyse the modality, as well as the quality of biometric samples. In many situations there are multiple biometric samples of the same modality available, or there are samples of different modalities available. In such situations the analysis includes some form of data fusion in order to optimize the accuracy of the biometric analysis. Biometric data fusion can be performed at different levels of the analysis and decision making process. This paper describes how subjective logic can be applied for biometric data fusion for multiple samples, multiple modalities and at different fusion levels.