Target Tracking And Trajectory Evaluation System Based On Radar Observation

Multi-target-tracking is a processing data which has a significant and important practical use for the military as well as the civilian surveillance applications.The strategic goal of the Multi-target-tracking is the tracking of the targets from an obtained data,this is what is especially complex because of many causes such as the variable number of the targets,the changing number of those moving targets,the noise,the wrong alarm and the uncertainty that contain the data association.To remain those problems,the most adequate solution is the use of the PHD filter(Probability Hypothesis Density),various estimations of this filter are presented in this thesis.CPHD(Cardinalized Probability Hypothesis Density)filter is also considered as one of the solutions of the defined problems but computationally more convoluted than PHD filter,we have also proposed an estimation of this filter in our thesis.In an engineer point of view,SMC-PHD filter which we are using in this thesis is as the most suitable filter for practical applications.Because of its ability to associate with the nonlinear and the non-Gaussian dynamics.Even though SMC-PHD generates some issues when extracting target State estimate,it is generally considered and still the most suitable approximation of the PHD filter.Four main methods are applied in this thesis(ZHAO method,Ristic method,the MEAP and the MEAP adaptive)in addition ‘the kmeans'(used for the comparison).The comparison of methods that we have used is demonstrated with the OSPA(optimal sub-pattern assignment)which is a new metric based on Wasserstein structure,but it completely eliminates most of the issues of the OMAT and the other metrics.This metric was presented only recently to solve some problems especially in process theory and spatial statistics.we have used OSPA in order to have the best estimation and filter evaluation,those techniques are totally different from the classical methods like(MHT(Multiple Hypotheses Tracker),JPDAF(Joint Probabilistic and Data Association Filter)).The first chapter is about the filters,the second is about the Multi-target state estimation,third about the new metric(OSPA)and last one is about the GUI application.Finally,in our thesis we have computed all the above methods to illustrate the comparison simulation and in addition we have also done a simulation of radar,we have linked the results obtained with the radar calculations to show a practical situation.