EKF和PF算法及其在目标跟踪中的应用

EKF and PF algorithm and their applications to target tracking

基于贝叶斯滤波的目标跟踪原理,介绍了扩展卡尔曼滤波(Extended Kalman Filter,EKF)和粒子滤波(ParticleFilter,PF)的基本思想和算法实现步骤。在非线性环境下对比分析了EKF算法和PF算法的估计精度,并给出两种方法的适用条件。EKF算法采用Taylor展开的线性变换来近似非线性模型,而PF算法采用一些带有权值的随机样本来表示所需要的后验概率密度。仿真结果表明,在强非线性非高斯环境下,PF算法的跟踪性能远优于EKF算法,当系统非线性强度不大时,EKF算法和PF算法的估计精度相差不大,但PF算法计算复杂,跟踪时间长,实时性差。

Based on the principle of Bayesian filtering theory, the basic idea and algorithm description of Extended Kalman Filter （EKF） and particle filter （PF） are introduced. Also, the estimation accuracies of EKF and PF are analyzed and the applicative conditions of the two different methods are given. EKF is a linearization technique which uses linear transformation of first order Taylor series expansion to approximate the nonlinear model. Particle filter represents the required posterior probability density by discrete random measures which are composed of weighted particles. The ex- perimental results demonstrate that the PF approach outperforms the EKF algorithm under strong nonlinear and non-Gaussian environment, while PF algorithm and EKF algorithm have the same tracking performance under weak nonlinear and non-Gaussian environment, but the PF is suffers from high computation complexity.