Algorithm Of The PHD Filter Based On Track-before-detect And Resampling Technology Research

With the rapid development of information technology,both in the field of military and civilian areas,multi-target tracking technology has a wide range of applications.In the current complex electromagnetic environment,the traditional algorithm for the weak target tracking still has some limitations,thus,researching better performance of multi-target tracking algorithm is imperative.In the framework of Bayesian estimation,one of the popular multi-target tracking algorithms,Probability Hypothesis Density(PHD)filtering theory is studied.At the same time,combining the advantages of Track-Before-Detect(TBD)to weak targets,this thesis focuses on the theory and implementation of PHD-TBD algorithm.In the traditional PHD-TBD filtering algorithm,there is no reasonable scheme for the newborn particle generation and the basic assumption of using PHD filter is not satisfied,so the performance and the practical value of the algorithm are severely limited.To combat these problems,an improved PHD filter based on TBD algorithm is proposed in this paper.First,an adaptive particle generation method based on differential localization is used.The new particles are gathered around the true target position,which greatly increases the effectiveness of new particles.Then,by establishing a new measurement,and using a small threshold to preprocess the observation data,the number of clutter approximately obeys the Poisson distribution,so that the PHD filter can better achieve its advantages in the TBD technology.The simulation results show that the proposed algorithm can improve the accuracy of target number estimation,enhance the detection and tracking performance,and reduce the computational cost.In TBD algorithm,since the amount of data in each frame to be processed is very large,in order to further improve the real-time performance of the algorithm,the resampling process in the implementation is mainly researched,and a buffered Metropolis Hasting(Buffered Metropolis Hasting,BMH)is proposed,which can be used in parallel pipeline operation.When states of particles and their weight are obtained in each update stage,the resampling operation can be started without having to wait for all the particles to be generated,and the particles with larger weights can be preserved in large quantities.The simulation results show that the BMH sampling method can achieve better tracking performance compared with the traditional method in the same multi-objective simulation scene,and further enhance the real-time performance of the algorithm.