摘要
针对闪烁噪声采用高斯分布建模敏感度大的问题,提出一种闪烁噪声下基于变分贝叶斯(VB)的扩展目标跟踪算法。在量测噪声逆协方差未知条件下,该算法首先将量测噪声建模为t分布,继而通过VB近似量测噪声逆协方差、目标状态以及自由度的后验概率密度,最后给出其高斯伽马混合分布实现。仿真实验表明,所提算法可以自适应地跟踪多参数未知情况下的多扩展目标,与传统算法相比具有较高的跟踪精度。
In view of the great sensitivity that gaussian distribution is modeled by glint noise, a new extended target tracking algorithm based on variational Bayesian with glint noise is proposed. With unknown measurement noise inverse covariance, t distribution is modeled by measurements, variational Bayesian method is applied to approximate posterior probability density of measurement noise covariance,target state and freedom. Then gaussian gamma implementation is given. Simulation results shows that the proposed algorithm can adapatively track multiple extended targets with uncertain parameters, while achieving better precision compared against the traditional approach.
出处
《商洛学院学报》
2016年第2期19-24,共6页
Journal of Shangluo University
基金
商洛学院科研基金项目(14SKY001)
关键词
多扩展目标
闪烁噪声
变分贝叶斯
目标跟踪
multiple extended target
glint noise
variational Bayesian
target tracking
