一种实用的融合方法在辅助粒子滤波中的应用

研究机动目标跟踪优化问题,传统的扩展卡尔曼滤波(Extended Kalman Filtering,EKF),无迹卡尔曼滤波(UnscentedKalman Filtering,UKF)跟踪机动目标强非线性系统,滤波精度有限,并存在噪声。为了提高跟踪滤波的精准性和可靠性,提出建立一种协同转弯非线性模型,将辅助粒子滤波(Auxiliary Particle Filtering)和数据融合相结合的跟踪滤波算法。辅助粒子滤波在重采样前对各个粒子权值依据似然函数进行修正,同时采用一种将多个滤波器结合的实用的数据融合方法,可以进一步提高滤波精度。经仿真表明,算法能减小运算量,并能很好的提高跟踪滤波精度。

Practical Data Fusion Algorithm Applied in Auxiliary Particle Filtering

The filtering for target tracking was discussed in the paper in order to find a method with better accuracy and reliability than extended kalman filtering and unscented kalman filtering.This paper built a non-linear system model,turn model.Then a method of auxiliary particle filters combined with data fusion was proposed.Auxiliary particle filtering for character of the turn model modified every particle before resampling according to likelihood function,which can get the certain accuracy of filtering with fewest particles.In order to improve the filtering result further,the multi-auxiliary particle filters with a practical data fusion algorithm was used and it is efficient in turn model and can get good result through computer simulation.