Dynamic particle allocation for CB-MeMBer filter

In this paper, we propose a particle allocation approach in the particle CB-MeMBer Filter for multi-target tracking (MTT). Considering the particle distribution uncertainty and existence probability of each target, we combine the re-sampling step of particle CB-MeMBer filter with the proposed particle allocation, resulting in the theoretically minimum multi-target filtering distortion for the system. Furthermore, we design a hardware structure of the proposed particle allocation approach and discuss how to optimize the resource usage in it. Simulation results show that particle CB-MeMBer filter with the proposed particle allocation approach has better filtering performance than that of the equally allocating approach, especially when the total number of particles is limited.