基于一种改进粒子滤波算法的目标跟踪研究

为了解决非线性、非高斯系统目标跟踪问题,研究了一种新的滤波方法——高斯粒子滤波算法。通过基于重要性采样和蒙特卡罗模拟方法得到一高斯分布来近似未知状态变量的后验分布。并讨论了此算法在机动目标非线性转弯运动中的跟踪应用,与粒子滤波算法相比,其优点是不需要重采样步骤。在闪烁噪声下比较了高斯粒子滤波器、粒子滤波器和扩展卡尔曼滤波器在滤波精度、运算时间等方面的差异,仿真结果表明该算法性能优于其他算法。

Research on Target Tracking Based on Improved Particle Filter Algorithm

A new Gaussian particle filter is introduced to solve the target tracking problem in the nonlinear and non-Gaussian systems.A single Gaussian distribution is obtained to approximate the posterior distribution of state parameters based on the sequential importance sampling and Monte Carlo method.The algorithm with its application in tracking the nonlinear and maneuvering targets is discussed.Compared with the particle filter,the resampling step is avoided.The accuracy,computational load and other aspects are compared with the Gaussian particle filter,the generic particle filter and the extended Kalman filter based on the glint noise statistic model.Results from the Monte Carlo simulation show that the performance of the GPF is superior to that of other filters.