Research On Multiple Extended Targets Tracking Based On Gaussian Process Regression

Because of the improvement of the radar and sensors resolution with the progress of science and technology constantly,target may generate more than one measurement in a moment,and an extended target is available.If still using the traditional method to associated the measurement and the target has been unable to meet the requirements.In recent years,the multiple target tracking method based on random finite set attracts the more attention of scholars due to its effective compared with the traditional algorithm.This thesis mainly focuses on the extended target tracking algorithm based on Gaussian Process Regression.The main research contents are as follows:1.In view of the complexity of estimating the shape of extended targets and the low accuracy in multiple extended target tracking in the clutters and missed detections,a Gamma Gaussian-mixture cardinalized probability hypothesis density filter with Gaussian Process Regression which can adaptively estimate the shape of the extended targets is proposed.Firstly,the extension of targets is modeled as a star-convex model,and on the basis of good estimation performance for motion state with Gamma Gaussian-mixture cardinalized probability hypothesis density filter,the Gaussian Process Regression is used to estimate the shape for extended targets,and thus achieve the purpose of tracking extended target.The simulation shows that the proposed algorithm outperforms the Gamma Gaussian-mixture cardinalized probability hypothesis density filter based on the star convex random hypersurface model in estimation precision and computing speed.2.The basic theory of maneuvering extended target tracking is studied,which includes the object dynamics model and the interacting multiple-model algorithm.Then,combined with the interacting multiple-model approach,an extended targets tracking algorithm with Gaussian Process Regression is proposed,in order to estimate the kinematic states and the extensions of a maneuvering extended target simultaneously.Firstly,based on Gaussian mixture probability hypothesis density algorithm,the interacting multiple-model algorithm is introduced to solve motor problem of extended target.Then,the Gaussian Process Regression is used to estimate the extended state of the maneuvering extended target,and the estimation precision of the extended state is improved by normalized processing for the measurement.The simulation experiments show that the proposed algorithm can improve the information loss caused by elliptic approximation and estimate the target's kinematic state,the extension and the classified state in real time.