改进的证据组合规则及其在融合目标识别中的应用

针对D-S理论在解决证据冲突时的局限性,提出了一种改进的证据组合规则.该方法充分考虑了证据源信息和冲突证据本身的信息,利用证据间的支持度矩阵,得到各证据的可信度.并以此对证据进行加权平均;在此基础上,将支持证据冲突的概率按各个命题在所有命题中所占的比重进行分配.仿真实验结果表明,该方法在处理冲突证据问题及证据比较一致的问题时,都能得到合理的融合结果.在实际的多传感器融合目标识别中,可以充分利用多个传感器的互补信息,最大程度地降低矛盾冲突因子对识别结果的影响,从而提高了融合识别的可靠性和有效性.

Improved combination rule of evidence theory and its application in fused target recognition

In view of the limits of D-S theory in dealing with conflict evidence,an improved evidence combination rule is proposed.In the method,the evidential source information and the conflict evidence distribution problem are well-considered.First,the reliability of each evidence is obtained by using the support-level matrix of evidences,and the evidence is weighted-averaged.Then,its conflict probability is assigned to each proposition according to its possession rate in all propositions.Numerical experiment shows...