A New Multi-source Information Fusion Method Based on Belief Divergence Measure and the Negation of Basic Probability Assignment

Abstract

Dempster-Shafer theory (DST) can effectively distinguish between imprecise information and unknown information, which is widely used in information fusion. However, when the evidence highly contradicts each other, it may lead to counter-intuitive results. In addition, the existing information fusion methods do not take the negation of BPA into consideration, which can be improved. In this paper, we propose a new information fusion method by taking into account not only the information in basic probability assignment (BPA) but also the information contained in the negation of BPA. In the method, the belief divergence measure is not only used to calculate the difference between BPA and its negative BPA to reflect the information volume carried by its initial BPA, but also to calculate the difference between BPA and other BPA to consider the discrepancy between evidence. The efficiency of the method is verified by case studies.