Novel biometric features fusion method based on possibility theory

摘要

In this paper, we propose a novel biometric modalities fusion method based on possibility theory. We have integrated this method in a bimodal biometric system. This biometric system is based on the fingerprint and the iris modalities to identify a person. The process of this method consists of two main phases: the features selection phase, and the fusion phase. In the first phase, we select the most relevant features by a proposed selection method based on the genetic algorithm and the possibility theory. Then, in the second phase, the selected features of different biometric modalities are fused by the novel proposed fusion method based on the possibility theory. An experimental evaluation of the bimodal biometric system which used the proposed fusion method shows the good performance of the possibility reasoning in the biometric features fusion.