Improved two-dimensional measure for evidence conflict

It is well known that D-S evidence theory can efficiently represent random and fuzzy uncertain information and combine evidence without prior information, therefore it can be used in many fields of information fusion. However, the combination result are always unsatisfactory when there are serious conflicts in evidences. To solve this problem, more researches have been done to improve the DS combination rule and modify the data model. However, the research of conflict is seldomly reported. Firstly, the weakness of classical conflict coefficient in D-S evidence theory was analyzed. Then, a distance function of evidence was introduced to measure conflict degree together with classical conflict coefficient. A improved two-dimensional conflict measure was proposed to describe evidence conflicts. Based on the new proposed conflict function, the applicability of Dempster's rule was analyzed. Numerical example shows that the proposed measure can efficiently determine the degree of evidence conflict.