An Intelligent Particle Filter Based Joint Probabilistic Data Association Algorithm for Long-Duration Multi-Target Tracking in Clutter Environment

The Joint Probabilistic Data Association Algorithm based on Particle Filter (JPDA-PF) is used in the tracking problem of nonlinear multi-target systems in dense clutter environments. However, during long time tracking, the tracking accuracy decreases with the increase of time. Meanwhile JPDA produces a combinatorial explosion in the number of associated events as the number of targets and clutter increases, greatly increasing the computational task. In this paper, a Joint Probabilistic Data Association Algorithm based on dolphin swarm particle filter (JPDA-DSPF) is proposed. Using the dolphin swarm optimization algorithm as a resampling method for particle filter can optimize the particle distribution to be closer to the region of high likelihood, and enhance the particle diversity to weaken the particle degradation phenomenon. We use an adaptive sequential approach to reduce computational costs by discarding correlated events with small probabilities. Finally, it is proved by simulation experiments that the method can effectively reduce the tracking error, inhibit the increasing trend of the error with time, effectively reduce the computation task and enhance the timeliness during multi-target encounters.