An Evidence Fusion Method Using Generalized Mahalanobis Distance in Dempster-Shafer Theory

For decades, Dempster-Shafer evidence theory provides a solution for information fusion with reduced uncertainty and ambiguity. However, it commonly has to suffer from the counter-intuitive result. This paper presents a new approach by combining the generalized Mahalanobis distance with the theory, which assures the computation of evidences' dissimilarities be calculable. The experiment demonstrates that the proposed method outperforms other techniques with a more reasonable aggregation.