目标属性信息相关时融合识别的实现方法  

Method of Implementation of Fusing Correlated Attribute Informationin Target Recognition

作　　者：王永成[1,2] 王宏飞[3] 杨成梧[3] 

机构地区：[1]郑州航空工业管理学院 [2]南京理工大学动力工程学院,江苏南京210094 [3]南京理工大学动力工程学院

出　　处：《兵工学报》2005年第3期338-342,共5页Acta Armamentarii

基　　金：河南省自然科学基金资助 (0 2 110 5 0 5 0 0 )

摘　　要：多源信息融合的一个主要应用方向是目标识别,但是各传感器测量到的目标属性值常常是相关的、不独立的,因此对各属性信息的融合比较困难。基于多属性决策模型(MADM )和D S证据理论提出了加权融合的方法,并给出了权重系数的计算方法和权重灵敏性分析方法,介绍了利用条件概率加权进行目标识别的方法。Target recognition is an important branch of application in multi-source information fusion. The attribute informations measured by each sensor are often correlated and dependent and therefore it is frequently a hard task to fuse data from different sensors. A novel weighted combination method based on the MADM and D-S evidence theory was put forward and the methods of establishing weight coefficients and analyzing their sensitivity were given in this paper. Moreover, the weighted conditional probability is also introduced into target recognition.

关 键 词：控制理论 信息融合 目标识别 多属性决策模型 证据理论 

分 类 号：N945.14[自然科学总论—系统科学]