基于SCKF的GM-δ-GLMB多目标跟踪算法

SCKF Based GM-δ-GLMB Multi-target Tracking Algorithm

多目标跟踪场景中,目标状态和量测均为随机分布。以高斯混合δ广义标签多目标多伯努利分布(gaussian mixtureδ-generalized labeled multi-bernoulli filter,GM-δ-GLMB)为代表的多目标跟踪方法,将状态和量测使用多目标多伯努利分量表示,通过多伯努利分量递推,实现对目标的跟踪与估计。GM-δ-GLMB在非线性多目标跟踪场景下会出现跟踪性能下降的问题。针对这一问题,将均方根容积卡尔曼与GM-δ-GLMB相结合,提高了GM-δ-GLMB算法在非线性场景的跟踪精度。同时,为减少运算复杂度和杂波对估计结果的影响,采用统计门限和最大似然概率策略,获取候选量测,用于多目标分量的更新。仿真结果表明,所提出的方法在非线性跟踪场景下具有良好的估计精度。

In multi-target tracking scene,target states and measurements are randomly distributed.In the multi-target tracking method represented by Gaussian mixture generalized label multi-target multi Bernoulli distribution(Gaussian Mixtureδ-Generalized Labeled Multi-Bernoulli Filter,GM-δ-GLMB),the state and measurement are represented by multi-target multi Bernoulli components,and the target tracking and estimation are realized by multi Bernoulli component recursion.However,GM-δ-GLMB has the problem of tracking performance degradation in nonlinear multi-target tracking scenarios.To solve this problem,the root mean square cubature Kalman algorithm is combined with GM-δ-GLMB algorithm to improve the tracking accuracy of GMδ-glmb algorithm in nonlinear scenarios.At the same time,in order to reduce the computational complexity and the influence of clutter on the estimation results,the statistical threshold and maximum likelihood probability strategy are utilized to obtain candidate measurements for multi-target component updating.The simulation results show that the proposed method has good estimation accuracy in nonlinear tracking scenarios.