The comparison of PHD algorithms with adaptive target birth

Probability hypothesis density (PHD) filter has received extensive attention because of its good performance in target tracking. Considering engineering practicability and traditional PHD filter implementation, several commonly adaptive PHD algorithms for newborn target intensity have been presented and analyzed in this paper. The comparative analysis including two categories filters and both of them are used for adapting newborn target intensity: measurement-driven extended PHD filter under sequential Monte Carlo (SMC) framework and measurement-gate partition based PHD filter under Gaussian mixture (GM) framework. The performance of different algorithms is verified from the aspects of state estimation precision, target number estimation and simulation time, which provides technical support for engineering applications.