用基于矩阵的D-S证据理论方法判定目标属性

Target attribution estimation by using matrix-based D-S evidence theory

D-S证据理论在多传感器数据融合中具有广泛的应用前景。其中Dempster综合规则可对可信度函数进行融合计算,当目标命题的个数很多时,融合计算容易造成漏项,引起计算错误。针对该问题,提出建立关于目标命题的关系矩阵以计算各目标命题的可信度分配值。通过关系矩阵,就可以利用计算机和Matlab语言编程实现自动计算,自动遍历所有项,避免漏项,保证结果的正确性。最后给出算法示例和应用Matlab程序计算的结果。

D - S evidence theory has found wide applications in the field of multi - sensor data fusion. Dempster rule can be used for fusing confidence - level function, but if the quantity of target subjects is very large, some items are liable to be omitted, which may result in mistakes in computation. In order to solve this problem, a relationship matrix about target subjects is set up to calculate the assigned confidence- level value of each subject. Auto computation is realized in Matlab by using relationship matrix so that all the items are considered and the validity of result can be ensured. An example based on relationship matrix and the calculation results are also provided.