基于量化测量的前向-后向箱粒子平滑器

Forward-backward box smoothing with quantized measurements

提出了一种新的前向-后向箱粒子平滑算法。在正向过程中,使用标准的箱粒子滤波算法。在后向过程中,使用箱粒子近似平滑后验概率。提出了一种额外的箱粒子移动步骤,使箱粒子在目标周围集中。通过量化测量下的仿真,验证了所提算法的性能优势。

In this work, we are interested in promoting tracking accuracy of target in quantized measurements by taking the de-laying measurements into consideration. Specifically, a novel forward-backward box particle smoothing algorithm is proposed. In the forward pass, the algorithm iterates with standard box particle filter. In the backward pass, smoothing posterior density is approxi-mated by box particles. An extra moving step of box particles is proposed to concentrate the box particles around the target. Simu-lations under quantized measurements are presented to verify the accuracy promotion of proposed algorithm.