目标综合识别系统中的多级分层属性融合方法研究  

Research on Multilevel and Delaminating Attribute Fusion Method for Target Synthetic Identification System

作　　者：王成刚[1] 

机构地区：[1]中国西南电子技术研究所,成都610036

出　　处：《舰船电子工程》2010年第5期67-69,154,共4页Ship Electronic Engineering

基　　金："十一五"国防预先研究课题项目资助

摘　　要：根据目标属性信息的层次化特征,提出了一种用于目标综合识别系统的多级分层属性融合方法。依据目标属性信息的内在关系,对其进行分类和分级,然后基于由高级属性推导低级属性的原理对由各种信源获得的属性信息进行预处理,最后使用DSmT的规则对同级属性信息进行融合。计算示例表明,该方法可以融合由异类信源获得的不同等级上目标属性信息,并且在高冲突的证据条件下依然有效。This article presents a new method based on hierarchical character of target attribute information to fuse multilevel and delaminating attribute for target synthetic identification system.Attribute information of target was classified and graded according to its internal relations.Then pre-processing of multi-source attribute information can be done based on the principle that low level attribute can be analysis on high level attribute.At last,the same level attribute can be fused according to DSmT rule.The analysis indicated that,different levels of attribute information got from different information source can be fused correctly by this method,and it still work with high conflict evidence.

关 键 词：综合识别 属性融合 多级分层 DSMT 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]