Research On Extended Targets Tracking With Nonlinear Measurement

Multi-target tracking technology is a research focus and difficulty in information fusion.Because of its high application value both in military and civil areas,multi-target tracking technology is the great attention of scholars both at home and abroad.With the continuous improvement of radar resolution,a target that generates multiple measurements at each time step is called the extended target.The Multi-target tracking method based on random finite set attracts the more attention of scholars due to its effective compared with the traditional algorithm.The random matrix and random hypersurface model are the two main approaches of modeling the extended states.This thesis mainly focuses on the extended target tracking algorithm with different measurement models.The main research contents are as follows:Firstly,aiming at the single extended target,the algorithm of single extended target tracking is given with the different models,such as random matrix model,random hypersurface model and multiplicative noise model.Secondly,due to that the measurement of the random matrix is linear,a method is proposed based on random matrix with nonlinear measurement.The algorithm deals with the nonlinear measurement of the extended target by the matched linearization,and then the linearized measurement equation is assumed as the measurement model of the random matrix.Aiming at the non-ellipsoidal extended object,a method is proposed based on random matrix with nonlinear measurement,which models the target as multiple ellipses,and then tracks it by the method of the random matrix.Finally,the random hypersurface model of the extended target is studied.For the nonlinear measurement model of the random hypersurface model,the extended target tracking algorithm based on the random hypersurface model is proposed with two different linearized methods,which are the stochastic model approximation and uncorrelated conversion filter.In view of the low accuracy in multiple extended target tracking in the clutters and missed detections of the traditional random hypersurface model,an algorithm based on Gaussian Process Regression is proposed.Because the measurement noise is unknown,a Multi-target tracking algorithm based on Gaussian Process Regression is presented,combining with the Variational Bayesian Expectation Maximization procedure.