Abstract: The conventional bootstrap filter suffers a main drawback of not incorporating the latest observations. Therefore, this paper proposed a Kalman Particle Filter (KPF) algorithm, and applied this new algorithm to bearingonly target tracking. An improved scheme was presented to handle this problem and yield a Kalman particle filter. The underlying idea of the new algorithm is that each particle is updated using Kalman filter incorporating the coming observations. A bearingonly tracking model was experimented and compared with bootstrap filter and KPF. The experimental results verify its superiority.

Key words: particle filter, Kalman filter, target motion analysis, linear tracking system

摘要： 针对基本粒子滤波算法没有融合当前时刻观测值的缺点，提出了一种卡尔曼粒子滤波算法。该算法针对每一个粒子使用卡尔曼滤波器进行更新，在更新过程中融合最新的观测信息，提高粒子滤波器的估计精度。针对纯方位目标跟踪问题进行实验，与基本粒子滤波算法及卡尔曼滤波进行了对比。实验结果表明，卡尔曼粒子滤波算法的跟踪性能明显优于其他两种算法。

关键词: 粒子滤波, 卡尔曼滤波器, 目标运动分析, 线性跟踪系统