自适应采样数粒子滤波算法

Adaptive Samples Particle Filtering Algorithm

基于统计决策规则提出自适应采样数粒子滤波算法,在定义综合性能风险函数的基础上,推导出粒子数与滤波误差方差之间的关系式,使得在跟踪过程中,可以根据目标的机动情况在线调节粒子数,以使跟踪性能达到最优。在Matlab仿真平台下进行了闪烁噪声下的机动目标跟踪实验,结果表明,自适应采样数粒子滤波算法是一种有效的机动目标跟踪方法,跟踪性能较基本粒子滤波算法提高了3.7倍。

This paper proposes an adaptive samples particle filtering（ASPF） algorithm based-on statistical decision-making rule, through defining the integrated performance risk function, deduces the relationship between the number of particles and filtering error variance, so the number of used particles during tracking can change according to the target maneuvering situation, in order to get the best possible tracking performance. Then, this paper demonstrates the performance benefits in an application of tacking a maneu- vering target in the glint noise environment, and the simulation results show that ASPF is an effective tracking method which makes the particle filtering performance raise 3.7times.