Multitarget Track-Before-Detect Based on Auxiliary Parallel Partition Particle PHD Filter

For dense multi-target scenarios, the existing Track-Before-Detect (TBD) algorithm based on Probability Hypothesis Density (PHD) has the shortcomings of underestimation of the number of targets and the waste of a large number of particles. The paper introduces the concept of two-layer particles and combines the Auxiliary Particle Filter (APF) PHD filter with the Parallel Partition (PP) theory to improve the estimation accuracy of the targets' number and state. The simulation results have shown that, compared with the existing PF-PHD-TBD algorithm, the newly proposed algorithm has a significant improvement in the estimation of the targets' number and the state, especially in dense multi-target scenarios.