扩展的证据合成规则及其在多目标跟踪中的应用  

Extended Evidence Combination Rule and Its Application on Multi-target Tracking

作　　者：刘宗香[1] 王品[1] 范恩[1] 

机构地区：[1]深圳大学信息工程学院ATR实验室,深圳518060

出　　处：《系统仿真学报》2013年第1期99-103,共5页Journal of System Simulation

基　　金：国防科技重点实验室基金资助课题(9140C8004011007)

摘　　要：针对多传感器多目标跟踪问题,利用随机集理论对证据合成规则进行了扩展,并基于"OR"合成规则提出了一种多传感器多目标跟踪的新方法。首先,将观测数据作为证据,目标状态作为识别对象,依据证据与识别对象间的相关度确立它们间的基本概率赋值。然后利用"OR"证据合成规则对证据进行合成。最后将合成的基本概率赋值作为权重因子对目标状态进行更新。仿真实验表明,与全局最近邻域方法相比,新方法能提高目标状态估计的精度。To solve the problem of the multisensor-multitarget tracking, the extended evidence combination rules were formulated in accordance with the random set theory, and a novel multisensor-multitarget tracking method based on the OR combination rule was presented. Firstly, with measurements as evidences and target states as identification objects, the basic probability assignments between the evidences and the identification objects were determined upon the association degrees between them. Secondly, the evidences were combined through the OR combination rule. Finally, with the combined basic probability assignments as weighted coefficients, the target states were updated. Simulation results show that the proposed method achieves higher state estimation accuracy than the GNN method.

关 键 词：目标跟踪 随机集理论 D—S证据理论 状态估计 

分 类 号：TN941.1[电子电信—信号与信息处理]