基于EK算法改进的多目标跟踪技术

Multitarget Tracking Technique Based on Extended Kalman Algorithm

在随机有限集理论框架下,传统的3种基于高斯混合实现的滤波算法仅适用于线性高斯系统中的目标跟踪。为了将算法的使用范围扩展到非线性高斯系统,文中提出了传统3种滤波器基于扩展卡尔曼(EK)改进后的算法,通过泰勒级数展开将非线性问题转化为线性问题近似处理,并对比分析了3种新算法对于匀速转弯目标的跟踪性能。仿真结果表明,3种新算法均能实现对于非线性高斯系统中目标的有效跟踪。

Under the framework of RFS theory,the traditional three filter algorithms based on Gaussian mixture was only applicable to linear Gaussian system.In order to extend to nonlinear Guassian system,three algorithms based on EK method were presented.Nonlinear problem was changed into linear problem approximatly by Taylor series expansion.Besides,the tracking performance of the three algorithms for targets with constant turning speed was compared.The simulation result indicated that the three algorithms can be applied to target tracking in nonlinear Gaussian system