Research On Multi-Target Detection And Tracking Based On Millimeter Wave Radar

Multi-target detection and tracking are important issues in many fields such as security monitoring,traffic control and intelligent driving.The millimeter wave radar has become one of the most promising detecting method due to its small size,light weight,high detection accuracy,excellent performances in harsh environments,and easy installation.The application background of this thesis is multi-lane traffic monitoring system.Based on the actual engineering project,combined theoretical analysis and simulation experiment with engineering verification,the detection and tracking of multi-target are studied with the wideband millimeter wave radar in LFMCW signal model.The key technology for multi-target detection is the separation of multiple targets,which aims to distinguish multi-target from three-dimensional information(range,velocity and angle).The resolution performance depends on the parameters of radar system and the algorithm.Therefore,in this thesis the millimeter wave radar system is firstly introduced,different ways of range and velocity decoupling with several LFMCW signal modulation modes is expounded,and the resolution problem in the multi-target condition is discussed.In addition,the sawtooth modulation scheme is applied after theoretical analysis and simulation experiments,and then the signal resolution capability is analyzed.In order to further distinguish targets from angular dimension,a variety of DOA estimation algorithms are compared after the separation processes of range and velocity.With fully considering the various algorithm features and multi-target application scenarios of this subject,3D-FFT algorithm is proposed,by which the high-resolution and high-precision target point-cloud distinguished from three-dimensional could be obtained.In order to track multi-target with three-dimensional resolution results,the basic selection principle of clustering algorithm is proposed firstly.According to the characteristics of target point-cloud,DBSCAN algorithm is applied for clustering.And then,the association algorithm of clustering results and target trajectories is studied.Aiming at the nonlinear problem between target motion model and radar system observation model,the extended Kalman filter algorithm is applied to track and filter the target motion state.In addition,the scheme of trajectory management is proposed.Finally,numerical simulations and experimental results demonstrate the performance advantages of proposed schemes.