一种学生t混合粒子实现的概率假设密度滤波器

Student’s t particle implementations of probability hypothesis density filters

在非线性多目标跟踪问题中,高斯混合粒子概率假设密度(GMP-PHD)滤波器在重尾的过程噪声和量测噪声的影响下会导致滤波性能的下降。针对该问题,提出一种新的学生t混合粒子概率假设密度(STMPPHD)滤波器。该滤波器将过程噪声和量测噪声近似为学生t分布,并用学生t混合模型来近似多目标的强度;同时,利用蒙特卡罗方法计算学生t积分,建立了学生t混合形式的闭式递推框架。仿真结果表明,该滤波器能够有效克服由重尾的过程噪声和量测噪声带来的不利影响,并能够保持较高的跟踪精度。

For the nonlinear multi-target tracking problem,the heavy-tailed process and measurement noises can reduce the performance of the Gaussian mixture particle probability hypothesis density( GMP-PHD) filter severely. To solve this problem,this paper proposed a new student’s t mixture particle probability hypothesis density filter( STMP-PHD). The method used a student’s t model to approximate the process noise and the measurement noise,and used a student’s t mixture model to approximate the intensity of the multi-target. The algorithm made the Monte Carlo method to calculate the student’s t integral,and establish the student’s t mixture closed recursive framework. The simulation results confirmed that the filter can effectively overcome the negative effects of the heavy-tailed process noise and the measurement noise,and maintain the high tracking precision.