Research On Robust Adaptive Irregular Shape Extended Target Tracking Method

Due to the widespread application of high-resolution detection devices and the demand for close-range target detection,extended target tracking problem has become a hot research topic in modern target tracking technology.In contrast to traditional point target tracking,ETT generates multiple measurements in a sampling period that contain a lot of characteristic information about target.The current core mission of ETT is to fully exploit the useful feature information in measurement set through information fusion techniques to obtain target shape contour and identify target in depth.However,it is challenging to obtain accurate and detailed feature information from measurement set,and the future development trend of extended target is how to scientifically and effectively process multiple sources of measurement information to make extended target state estimation more accurate.In view of this,this paper will carry out the study of robust adaptive irregular shape ETT method to explore a series of evaluation system of extended target shape performance and shape estimation theoretical methods.The main research contents of this thesis are as follows:(1)Aiming at the shape performance of extended target for irregular shape,an evaluation system based on Fréchet distance and area error is proposed to measure the similarity between actual shape and estimated shape of target.When area error is used to evaluate shape performance of ETT algorithm,there is a certain probability that the numerical results are the same but the shape is not similar.Therefore,Fréchet distance is introduced for adaptive weighting.Fréchet distance describes the similarity between actual shape and estimated shape using the spatial distance of curve by continuously reparameterizing the curve.First,Fréchet distance is normalized to make the comparison result simple and clear.Then,an adaptive weight coefficient is used to weight Fréchet distance and area error,and sign distance function is combined to reflect relative size relationship between actual shape and estimated shape.Finally,the proposed method and mainstream extended target shape performance evaluation index are constructed for simulation comparison experiments to verify the effectiveness of proposed method.(2)Aiming at the problem of irregular ETT in complex environment,a robust adaptive star-convex extended target tracking method based on Artificial Potential Field theory is proposed to jointly optimize kinematic state estimation and extended state estimation of extended target.First,under the theoretical framework of APF,a measured potential field is established with measurement information,and nonlinear measurement equation is further derived according to gradient information in the established measured potential field.Then,optimal nonlinear filter is combined to update target state,and updated extended state is represented by B-spline curve.Finally,root-mean-square error and Fréchet-Area Weighted distance are used as the evaluation indexes of motion performance and shape performance of star-convex extended target,respectively,and the simulation experiments of static tracking and dynamic tracking on the star-convex extended target with different shapes are carried out to verify the effectiveness of proposed method on irregular shape ETT.(3)Aiming at the problem of irregular ETT in complex environment,a robust adaptive non-star-convex extended target tracking method based on Level Set(LS)theory is proposed to optimize the state estimation of extended target.First,the shape function of filling shape is described according to LS theory,which is generally represented by sign distance function.Then,Gaussian Process is used to learn nonlinear mapping relationship between input and output to obtain a posterior probability distribution with the aim of finding boundary function maximum.The nonlinear measurement equations combining LS theory and Gaussian process are further derived to update the target state of non-star-convex extended target in the framework of optimal nonlinear filtering.Finally,Fréchet-Area Weighted distance is used as shape performance evaluation index of non-star-convex irregularly shaped extended target,and simulation experiments are constructed to verify the effectiveness of proposed method for non-star-convex shaped ETT estimation.