摘要
针对传统多目标概率假设密度(PHD)滤波器在低检测概率情况下跟踪精度低和失跟率高的问题,提出了一种改进的概率假设密度滤波算法。该算法利用高斯混合PHD(GM-PHD)滤波器进行PHD预测和PHD更新,处理过程中通过修正上一拍权值大的高斯项,并在处理当前拍时保证其权值的稳定性,以保证算法的高精度。仿真结果表明,在低检测概率情况下,该算法可较好估计目标数和目标状态。与传统GM-PHD滤波器比,该算法跟踪精度大幅提高。
To solve problems of low tracking accuracy and high probability of target tracking loss with low detection probability for the conventional probability hypothesis density（PHD）filter,an improved PHD（IPHD）filter algorithm is proposed.In the algorithm,the Gaussian mixture PHD（GM-PHD）filter is used to execute PHD prediction and PHD update.In the process,the high accuracy of the proposed algorithm is achieved by revising the Gaussian components with large weight at the previous one time step properly and promising the stability of the weights in the process of the current time step.Simulation results show that in the situation of low detection probability,the algorithm can estimate the target number and target states well compared with the traditional GM-PHD filter,and the tracking accuracy improves largely.
出处
《指挥信息系统与技术》
2014年第6期36-40,共5页
Command Information System and Technology
基金
国家自然科学基金(61374159
61374023
61203233
61203234
61135001)
航空科学基金(20125153027)资助项目
关键词
目标跟踪
低检测概率
概率假设密度(PHD)滤波器
高斯混合PHD滤波器
权值
修正
target tracking
low detection probability
probability hypothesis density （PHD） filter
Gaussian mixture PHD （GM-PHD） filter
weight
revising
