传感器网络中结合分类信息的多目标关联算法  

Multi-objective connected algorithm with category message in sensor networks

作　　者：付兴武[1] 高芳芳[1] 白风[1] 

机构地区：[1]辽宁工程技术大学电气与控制工程学院,辽宁葫芦岛125105

出　　处：《传感器与微系统》2010年第1期91-93,97,共4页Transducer and Microsystem Technologies

基　　金：国家自然科学基金资助项目(50874059);辽宁省科技攻关计划资助项目(2007231003)

摘　　要：提出一种基于图模型的结合目标分类信息数据关联算法,目的在于提高多传感器网络的数据关联正确率。通过分析基于图模型的目标分类信息,适当修改顶点兼容函数和连线兼容函数的计算公式,并用结合分类信息的似然函数更改原来只利用运动学信息的兼容函数,最终将分类信息应用到关联算法中。通过对仿真结果的比较,验证了利用分类信息可以较好地改善多目标关联性能。A data associaed algorithm based on graphic model which combined with target category message is proposed. The goal is to enhance data association accuracy of the multi-sensor networks. Through the analysis of target category information based on graphic model, the apex compatible function and the wired compatible function formula are suitably revised, and combining with category message＇ s likelihood function to change compatible function which originally use kinematics infoimation. The category message is applied to connected algorithm. Through comparison of simulation result, it is confirmed that category message can be used to improve the connected performance of multi-objective.

关 键 词：图模型 分类信息 数据关联 

分 类 号：TP212.9[自动化与计算机技术—检测技术与自动化装置]