Maneuvering Target Tracking Algorithm Based On Parameter-Adaptive Particle

Maneuvering target tracking has always been a difficult problem in the field of target tracking,especially tracking multiple maneuvering targets.Due to the change of the number of targets,the new and new targets disappear,and the complex scenes such as targets being close to each other and crossing each other make tracking more difficult.In response to this problem,under the filtering framework of random finite set,this paper focuses on the research of multimaneuvering target tracking algorithm with variable number and arbitrary motion in complex environment.The main work includes:1.Aiming at the problem of multi-target tracking when the target motion model is unknown and the new target intensity is unknown,this paper proposes an adaptive grid-driven filtering algorithm based on PHD and CPHD filtering.First adjust each target grid adaptively according to the target grid resolution,and then adjust the grid position according to the target measurement and status,in addition,the new target recognition technology is used to identify the new-born target to achieve adaptive filtering to achieve the unknown and new target movement Multi-target tracking in complex situations where the model is unknown.The experimental results show that the proposed algorithm can realize multi-target adaptive tracking of unknown motion parameters and unknown number of new targets.In terms of average OSPA distance and average target number estimation,it has better tracking performance than traditional particle PHD filtering methods.2.In the complex environment,the multi-target tracking problem with unknown new strength,unknown measurement noise and arbitrary maneuvering.In this paper,a parameter adaptive particle potential probability hypothesis density filtering algorithm is proposed.The adaptive parameter estimation technique is integrated into the CPHD filtering framework.First,the inverse gamma distribution is used to approximate the posterior distribution of noise,and then the adaptive LW filtering is used.Estimate time-varying maneuvering parameters.In addition,new-born target recognition technology is used to identify new-born target to solve the problem of arbitrary maneuvering multi-target tracking.Experimental results show that the newly proposed algorithm can correctly identify unknown measurement noise variance and can track multiple maneuvering targets in a more robust manner.3.For PHD and CPHD filtering,the target track cannot be obtained.Based on the parameter adaptive particle potential probability hypothesis density filtering,this paper proposes to introduce the track correlation algorithm to the parameter adaptive particle potential probability hypothesis density filtering algorithm.In addition,the parameter adaptive particle potential probability hypothesis density and grid-driven PHD/CPHD filter algorithm are compared and analyzed in maneuvering scenarios.Experimental results show that the newly proposed algorithm can obtain the track information of each target in complex scenarios such as maneuvering,and is not affected by the number of targets.The parameter adaptive particle potential probability assumes that the density algorithm has a low calculation time complexity.The grid-driven PHD/CPHD filter algorithm has a calculation time complexity and tracking error that are related to the number of grids.The more grids,the higher the time complexity and tracking.The smaller the error,in addition,the parameter adaptive particle potential probability assumes that the density algorithm is more affected by process noise than the grid PHD/CPHD filter algorithm.