Research On Information Fusion Method Based On Evidence Theory

The successful application of information fusion in military has attracted attention widly.With the continuous expansion of application,the increasing complexity of application background and the deepening of application level,the problem of uncertain information processing gradually highlights,becoming a common problem faced by information fusion system applied in most field.Evidence theory is favored by its unique advantages in the representation and reasoning of uncertain information.However,the application of evidence theory in information fusion is also confronted with the problem of the construction of basic probability assignment function and the counterintuitive results caused by conflict information ftusion occasionally.This dissertation studies these two aspects with a view to evidence theory can be better applied in information fusion system,hence improve the performance of information fusion.In view of the construction of the basic probability assignment function,a method of construction based on the fuzzy membership degree is proposed by virtue of the advantage of the fuzzy set in describing imprecise information.This method firstly determine the structure of focal element according to the intersection point of observation value and membership function,and then a method of focal element value assignment with focusing properties is proposed according to the intrinsic characteristics of membership degree distribution.The simulation results show that the basic probability assignment function constructed by this method has the advantages of good focusing property and accurate description of uncertain information,and also has the advantages of small conflict and fast convergence speed in the subsequent operation.In order to solve the problem of conflict information fusion,combining the support relationship between evidence and the uncertainty of the evidence itself,a method of evidence discount based on uncertainty is proposed,and two kinds of applications condition are given.On this basis,a method of conflict information fusion based on uncertainty is proposed.The simulation results show that the proposed method can guarantee the correctness and rationality of the fusion results,and has good robustness and convergence.Finally,the iris identification information fusion system was built based on the iris data set,the results show that the method proposed in this dissertation has a high correct recognition rate for iris category identification system,the basic probability assignment function constructed by the proposed method can describe the information of the system very well and has good operation characteristics.The proposed conflict information fusion method can guarantee the correct result when the system has conflict information,which improves the flaw of classical evidence theory in the fusion of conflict information.