基于边缘粒子滤波的目标跟踪算法研究

为了提高目标跟踪过程中粒子滤波结果的精度，将边缘粒子滤波算法应用于目标跟踪。首先将目标运动状态向量划分为线性和非线性两个子向量，然后，采用卡尔曼滤波方法处理线性状态子向量，采用粒子滤波方法处理非线性状态子向量。使用边缘粒子滤波算法和标准粒子滤波算法对目标进行跟踪仿真。仿真结果表明：将边缘粒子滤波算法应用在目标跟踪过程中，能够取得更高的跟踪精度；时间复杂度增加仅6%；在粒子数相对较少的条件下，仍能够保持较好的滤波性能。

Research on target tracking algorithm based on marginalized particle filter

In order to promote accuracy of Particle Filter（PF） in target tracking,Marginalized Particle Filter（MPF） is used.Firstly,state vector of target motion can be divided into linear state sub-vector and nonlinear state sub-vector.Secondly,Kalman filter is applied to the linear state sub-vector,and particle filter is applied to the nonlinear state sub-vector.Simulation is given by using standard PF and MPF to track a target.Results show that the MPF has some better characters.Such as,it can get higher accuracy tracking results,its increasing ratio of time complexity is only six percent,and it can maintain better performance even if there are a few particles contrastively.