基于证据理论的目标识别方法  

Target recognition method based on evidence theory

作　　者：张燕君[1] 龙呈[1] 

机构地区：[1]燕山大学信息科学与工程学院特种光纤与光纤传感河北省重点实验室,河北秦皇岛066004

出　　处：《系统工程与电子技术》2013年第12期2467-2470,共4页Systems Engineering and Electronics

基　　金：河北省科学技术研究与发展计划应用基础研究计划重点研究基础项目(No.11963545D)资助课题

摘　　要：当目标识别系统中传感器信息高度冲突时,仅利用D-S组合规则无法有效融合。提出一种基于证据理论的目标识别方法,该方法首先定义了冲突系数,在此基础上设计了目标识别方法。当冲突较小时,直接利用D-S组合规则进行融合识别;反之,根据冲突的具体情况对证据先折扣再融合识别。通过目标识别仿真实验与其他识别方法进行了比较,实验结果表明所提方法提高了抗干扰能力,具有较快的收敛速度。When the sensor information of the target recognition system is highly conflict, integration can not be effectively carried out only using the D-S combination rule. A target recognitive method based on evi-dence theory is presented. On the basis of defining a conflict coefficient, a target recognitive method is designed. When conflict is low, the D-S combination rule is used directly. On the contrary,according to the specific cir-cumstances of conflict evidence, the evidence is discounted firstly and then is fused and intergrated. Compared with other methods, simulation experimental results show that this proposed method can improve the anti-jam-ming ability and has a faster convergence speed.

关 键 词：信息融合 目标识别 证据理论 冲突系数 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]