Target Tracking With The Distrubuted Through-the-wall Radar

Distributed through-wall radar network can achieve the reliable detection,precise positioning and dynamic tracking of the human body by multiple detection nodes emit electromagnetic waves to see the building wall or the jungle foliage,which has extremely important potential for application in the fields of public safety.In this thesis,the time registration of the network,the registration of the space and the dynamic tracking after the target imaging are discussed in view of the gap of target tracking technology in the field of distributed through-wall radar network.It's as follows:1.For the application scenario and task attributes of distributed through-wall radar network,the basic form of distributed network are researched.The topology and working flow of distributed probe network are proposed.The ranging method is proposed with the bidirectional time,and the error compensation is introduced.Two kinds of nodes-self-localization methods are studied,and the ellipse-hyperbola joint-cross-localization algorithm is proposed.The simulation shows that it's better than traditional self-localization method.In this way,the structure construction and space-time registration of distributed radar network are realized.2.For the limitation of traditional single-view through-wall radar applications,three kinds of target motion models under complex environment are studied.Two types of target tracking algorithms are proposed for distributed through-wall radar network,that is,centralized and distributed.3.For the centralized target tracking architecture,the algorithm is designed.The feature distribution of the target image is studied,and a feature level image fusion algorithm based on the target feature information is proposed.The electromagnetic simulation results show that the signal-to-noise ratio of the fused target image is significantly higher than that of the conventional pixel-level fusion.The image clustering method is studied,and the radar image clustering algorithm based on feature weighting factor is proposed.The simulation shows that the algorithm can optimize the positioning result.The method of video image tracking is studied,and the radar image tracking algorithm based on probability mean shift is proposed to detect,locate and track the missing detection targets under single view.4.For the distributed target fusion architecture,the main function of the algorithm is designed.Based on the multi-model target tracking,and using the track management method,the tracking algorithm of the single-station through-wall radar is improved.The simulation validates tracking performance.The distributed trajectory fusion algorithm is studied.A trajectory matching based on weighted distance is proposed,and a covariance cross-fusion algorithm based on the least determinant solution is adopted to achieve all target detection.The results of processing the same measured data are compared and analyzed,and a practical scheme of combining two target tracking algorithms is proposed.