并行高斯粒子滤波被动多目标跟踪新算法

针对多目标跟踪的数据关联及多目标状态空间尺寸随目标数增多而增长的问题,提出了一种跟踪新算法,假定各目标的状态与过去的观测相互独立,可以多路并行处理,采用马尔可夫链蒙特卡罗(MCMC)方法计算目标与观测的关联概率,利用高斯粒子滤波(GPF)独立估计单个目标的状态,采用拟蒙特卡罗(QMC)方法近似各目标的预测及更新分布.将该算法应用于被动多传感器多目标跟踪,仿真结果表明,所提算法比联合概率数据关联滤波器(JPDAF)、马尔可夫链蒙特卡罗数据关联滤波(MCMCDAF)及蒙特卡罗联合概率数据关联滤波(MC-JPDAF)具有更好的跟踪性能.

Parallel Gaussian particle filter for passive multi-target tracking

In multi-target tracking,aiming at the data association problem and the curse of dimensionality that arises due to the increased size of the state-space associated with multiple targets,a novel algorithm based on the parallel Gaussian particle filter(parallel-GPF) is proposed.By assuming that the states of the targets conditioned on the past measurements are mutually independent,the algorithm is amenable for multi-channel parallel implementation.With the Markov chain Monte Carlo(MCMC) approach the marginal ...