一种改进重采样的粒子滤波算法

针对粒子滤波重采样过程中存在的粒子多样性丧失问题,提出一种改进重采样的粒子滤波算法。按照局部重采样算法对粒子进行分类,中等权值的粒子保持不变,大、小两种权值的粒子采用Thompson-Taylor算法进行随机线性组合产生新粒子。实验结果表明,该算法能在降低计算复杂度的同时不丧失粒子多样性,提高了滤波性能。

Particle filter algorithm based on improved resampling

In order to solve the loss of particle diversity exiting in resampling process of particle filter, this paper presented a particle filter algorithm based on improved resampling. It classified the particles to different groups according to partial resampling. It kept the particles with medium weight values same, and combined the other two groups with high and low weight values linearly and randomly to generate new particles using Thompson-Taylor algorithm. Experimental results show that the improved algorithm can reduce computational complexity and keep the diversity of particles and it also enhances the performance of filter.