| Due to the development of modern science and technology,radar system is becoming more and more complex,and radar is also required to have the ability to track multiple targets at the same time.In the application field of target tracking,the demand for multi-target tracking technology is becoming more and more stringent.In this dissertation,the key technologies of multi-target tracking in dense clutter environment are studied in depth,and the following parts are mainly studied.Firstly,in the dense clutter environment,the joint probabilistic data association multitarget tracking algorithm will increase with the increase of measurement information and the uncertainty of measurement sources,which leads to the increase of algorithm calculation and the decrease of algorithm real-time performance.This dissertation studies the improvement of joint probabilistic data association multi-target tracking algorithm.By redefined the confirmation matrix and linear approximation association probability,the possible matching number between measurement and target is reduced,the complexity of association probability calculation is reduced,and the real-time performance of the algorithm is improved.The simulation results show that,while ensuring the accuracy of multi-target tracking algorithm,the improved algorithm is twice as real-time as the classical data association algorithm in the target tracking process,which greatly improves the real-time performance of multi-target tracking algorithm based on joint probability in dense clutter environment.Secondly,aiming at the multi-target tracking problem when the number of targets is unknown in dense clutter environment,this dissertation studies the trajectory management mechanism based on sequential probability ratio test.By adding the trajectory management mechanism to judge the start and end of the target trajectory in the measurement,the statistics of the number of targets in the multi-target tracking process when the number of targets is unknown can be realized.At the same time,considering the real-time performance of the multitarget tracking algorithm when the number of targets is unknown,it is proposed to improve the joint probability data association algorithm by redefining the confirmation matrix and linear approximate association probability,and to combine it with the trajectory management mechanism based on the sequential probability ratio test.The simulation results show that the accuracy of the multi-target tracking algorithm is guaranteed,and the real-time performance of the algorithm is improved by tenfold.The multi-target tracking with unknown target number in dense clutter environment is realized.Finally,aiming at the problem of multi-target tracking when the target trajectory spacing is close or the target trajectory cross in dense clutter environment,this dissertation introduces clustering idea to realize target tracking,and studies the multi-target tracking algorithm based on Gaussian mixture model.Considering that the tracking accuracy of multi-target tracking algorithm based on Gaussian mixture model decreases when the target motion state becomes complex,a multi-target tracking algorithm based on feature selection is proposed.The target and clutter in the component measurement information of feature correlation index region are used.The position difference of measurement information is introduced as auxiliary feature to improve the influence of clutter on multi-target tracking.The simulation results show that the improved Gaussian mixture model multi-target tracking algorithm based on feature selection has certain advantages in tracking targets with close range motion,and its real-time performance and tracking accuracy are better than those based on joint probability. |
