基于改进高斯粒子滤波器的目标跟踪算法

针对现有机动目标跟踪中粒子滤波算法的不足，提出了一种改进的粒子滤波方法。该方法在高斯粒子滤波的基础上通过利用当前时刻量测值对量测误差的分布参数进行实时的统计和更新，并以此得到粒子的权值，从而考虑到了量测值对估计值的影响，该方法适合于量测误差分布为高斯白噪声且状态量与量测误差相关条件下的非线性估计。仿真结果表明，与传统的自举粒子滤波(boot trap particle filter, BPF)、高斯粒子滤波（Gaussian particle filter, GPF)以及无迹粒子滤波（unscented particle filter, UPF)相比，该方法具有较高的精度和较少的计算量。

Object tracking method based on improved Gaussian particle filter

An improved method is put forward based on Gaussian particle filter aiming at overcoming the disadvantages of existing particle filter methods. The current measures are taken into consideration to estimate and append the distribution parameters of measure errors for getting the weighted factors of particles. The method can be used to solve non linear problems which have conditions that measure error distribution is Gaussian and state values have relatives with measure errors. Simulation results show more accuracy and less computation cost comparing with traditional PF, UPF and GPF methods.