论文部分内容阅读
针对非线性、非高斯的跟踪模型,提出了一种新的多目标跟踪滤波算法,称为高斯和概率假设密度(Gaussian sum probability hypothesis density,GSPHD)滤波器。分析的结果表明,当初始的先验强度满足高斯或者可以表示成高斯和的形式时,通过将状态噪声、观测噪声、目标的繁衍、新目标的产生、目标的存活概率和检测概率表示成高斯和的形式,之后每个时刻的后验强度均能表示成高斯和形式。分析还表明,现有的混合高斯概率假设密度(Gaussian Mixture probability hypothesis density,GMPHD)滤波器是本文提出算法的一个特例,本文GSPHD弥补了现有GMPHD不能处理非高斯噪声模型的不足。目标跟踪的仿真结果表明了提出算法的有效性。
Aiming at nonlinear and non-Gaussian tracking model, a new multi-target tracking filtering algorithm is proposed, which is called Gaussian sum probability hypothesis density (GSPHD) filter. The results of the analysis show that when the initial a priori strength satisfies Gauss or can be expressed in the form of Gaussian sum, the survival probability and detection probability of the target are expressed as Gaussian by combining state noise, observation noise, propagation of the target, generation of new target, And the posterior strength at each moment can be expressed as Gaussian and formal. The analysis also shows that the existing Gaussian Mixture probability hypothesis density (GMPHD) filter is a special case of the algorithm proposed in this paper. In this paper, GSPHD makes up for the deficiency that the existing GMPHD can not handle the non-Gaussian noise model. Simulation results of target tracking show the effectiveness of the proposed algorithm.