基于重采样粒子滤波的目标跟踪算法研究

基于传统粒子滤波的机动目标跟踪方法针对非线性、非高斯系统有较好的估计性能,但是存在粒子退化现象。利用残差重采样算法,可以有效克服粒子滤波的退化问题。本文针对残差重采样算法作进一步研究,提出了一种改进的残差重采样粒子滤波算法。该方法在残差重采样基础上进行改进,可以避免残差重采样中关于残留粒子的重采样问题,在保证精度的前提下提高运行效率,减少运算复杂程度。仿真实验结果表明该算法与残差重采样粒子滤波相比提高了目标跟踪的实时性,并且随着粒子数的增加,这种优势表现得更加明显。

Research on Target Tracking Method Based on the Resampling Particle Filter

The method of target tracking based on traditional particle filter performs well when esti-mating non-linear/non-Gaussian systems , however particles degeneration can occur .The particles degen-eration can be overcome by using residual resampling algorithm .Through the research on residual resam-pling algorithm , an improved residual resampling particle filter algorithm is presented in this article .The method which improves the residual resampling algorithm can avoid the resampling of residual particles , thus improving running efficiency and reducing computational complexity .Simulation results show that the effect of real-time target tracking of this improved algorithm is higher than that of residual resampling par -ticle filter.When there are more particles , the efficiency is more outstanding .