摘要
针对传统粒子滤波目标跟踪算法存在的粒子退化问题,提出了一种新的基于支持向量机的粒子滤波目标跟踪算法。该算法利用滤波过程中的预测粒子集及其权值,使用支持向量机估计出系统状态的后验概率密度,再根据该概率密度重采样更新粒子集,以提高粒子的多样性,从而克服粒子的退化现象。仿真结果表明,该算法能显著增加有效粒子的数量,其目标跟踪精度优于马尔可夫链蒙特卡罗移动方法以及正则粒子滤波算法。
To solve the problems of particle degeneration in traditional particle filter,an improved target tracking algorithm was proposed based on density estimation with support vector machines.Using support vector machines,the posterior probability density function of the state was estimated with predicted particles and their important weights during filter iteration.After resampling the new particles from this density model,the degeneration of the filter was eliminated effectively by these diversiform particles.Simulation results demonstrate that the proposed algorithm can increase the quantity of effective particles obviously,and the new filter is superior to the Markov Chain Monte Carlo particle filter and regularized filter.
出处
《吉林大学学报（工学版）》
EI
CAS
CSCD
北大核心
2011年第4期1102-1106,共5页
Journal of Jilin University:Engineering and Technology Edition
基金
'十一五'国防预研项目(51309030102
51309030203)
关键词
自动控制技术
目标跟踪
粒子滤波
支持向量机
概率密度估计
automatic control technology
target tracking
particle filter
support vector machines
density estimation
