改进的粒子滤波算法在多目标跟踪中的应用研究

Research on Improved Particle Filtering Algorithms in the Multiple-Target Tracking

由于基于序贯重要性采样的粒子滤波算法存在着样本退化的问题,因此文章在几种常用的重采样算法的基础上提出了一种改进的重采样算法,通过在初始化阶段对粒子集的优化处理,在重采样阶段使用基于特定权值的改进重采样算法,从而得到了一种改进的粒子滤波算法。最后根据仿真实验表明改进的算法不但在跟踪精度上有所提高,而且对于样本退化和枯竭问题也进行了一定程度的改善,更为重要的是在多机动目标跟踪中也得到了很好的应用。

Because the problem of sarnple degeneracy exist in the SIS（Sequential Importance Sampling）particle filte- ring algorithms, an improved resampling algorithms has been presented based on several common resampling algorithms. The particle disjoint set were optimized in initialization stage, then an improved resampling algorithms were carried out based on specifically weights, and an improved particle filtering algorithms has been received. The emulation experiments, show that the improved algorithms not only improved the tracking accuracy, but also relieved the sample degeneracy and impoverishment in a certain degree. The more important thing is that it was applied very well in the multiple target tracking.