改进的粒子滤波算法

New Improved Particle Filter

针对标准粒子滤波算法重采样后粒子多样性丧失问题,提出一种在粒子补偿的基础上,利用预测值与观测值的方差进行粒子重采样的改进粒子滤波算法。该算法是在标准粒子滤波算法的基础上,加入粒子补偿的步骤,然后利用预测值与观测值的方差在权值较高的粒子周围进行重采样来改善标准粒子滤波算法中粒子多样性丧失问题。实验结果表明:在相同条件下,改进粒子滤波算法比标准粒子滤波算法具有更小的平均均方误差(RMSE)和更高的目标跟踪精度,数据表明,其目标跟踪精确度提高30%以上。

Point to the issue that after standard particle filter for resample, it would come into being problem of the loss of particle diversity. For this reason, a new improved particle filter algorithm is presented. The algorithm is based on standard particle filter, adding particle compensation steps, and changing the method of resample. The method is that particles resample around of the high wrights of particles by the variance of the predicted values and the observation data. The improved particle filter is compare for the standard particle filter, on the basis of the experimental results, improved algorithm has smaller average tnean -square errors （RMSE）, and more target tracking precision. The data show that the tareet trackinz orecision raises 30% above.