Research On Three-dimensional Imaging Method For Space Targets Based On Multi-view ISAR Image

Inverse synthetic aperture radar(ISAR)can acquire high resolution one-dimensional(1-D),two-dimensional(2-D)and three-dimensional(3-D)images of space targets in a long range throughout the day in all weather.Therefore,it is widely used in space situational awareness and battlefield situation reconnaissance,etc.However,since the 1-D and 2-D ISAR images are projections of the 3-D structure of the space target in the radar line of sight(LOS)and imaging planes,respectively,they are difficult to truly reflect the geometric characteristics of the target.Therefore,in recent years,more and more scholars have beginning to focus on the research of 3-D ISAR imaging.This dissertation mainly focuses on the research of 3-D ISAR imaging methods based on multi-view ISAR images for space targets,which currently faces two major difficulties remained to be solved.On one hand,due to the serious anisotropy of the scatterers in the ISAR images,it is difficult to stably track the positions of the scatterers in the ISAR image sequence,resulting in incomplete tracks of scatterers.Therefore,it is difficult to form a trajectory matrix and obtain the 3-D coordinates of scatterers via matrix decomposition method.On the other hand,for the space targets with extra unexpected self-motion,there is an error on the projection vectors calculated by the observed LOS.Hence,it is difficult to accurately describe the mapping relationship between the 3-D coordinates of the target scatterers and the 2-D projection positions,resulting in the projection mismatch.The above two challenges are the key problems in the research of 3-D ISAR imaging,and they are also the problems in practical application of ISAR 3D imaging technology.Therefore,the research in this dissertation has high theoretical significance and great application value.In this regard,combined with the urgent needs of 3-D ISAR imaging,this dissertation delves into the researches on trajectory association for scatterers in ISAR image sequence,3-D imaging for triaxial stable space targets and fixed-axis slowly rotating space targets,and3-D ISAR imaging for targets without motion prior.The specific innovations of this dissertation are as follows:1.Aiming at the problem that it is difficult to associate scatterers’ trajectories caused by complex scattering characteristics such as anisotropy of scatterers in ISAR image sequences,a trajectory association method based on random finite set(RFS)theorem is proposed.Firstly,the trajectory form of the scatterers on a space target in 2-D ISAR imaging planes is derived and analyzed in detail.Then,under the framework of random finite set theory,the states of target scatterers and measurements are modeled as random finite sets,and a Modified Labelled Gaussian Mixture Probability Hypothesis Density(ML-GM-PHD)filtering algorithm is proposed to achieve accurate association of scatterers’ trajectories.Then the 3-D imaging result of the space target is realized via the factorization method.The experiments on scatterers’ trajectories association and 3-D imaging of space targets based on the simulated data verify the effectiveness and robustness of the proposed method.2.Aiming at the problem that the 3-D imaging methods based on factorization can only obtain sparse 3-D point clouds of space targets,a novel 3-D ISAR imaging method based on image sequence energy accumulation(ISEA)is proposed.Firstly,according to the motion characteristics of different space targets,the motion models of the triaxial stable targets and the fixed-axis slowly rotating targets are established,respectively.Then,the mapping relationships between the 3-D coordinates of the target’s scatterers and the projection positions in the ISAR image sequence under different motions are derived.After that,by taking advantage of the fact that the projection positions of the real scatterers of the targets in each frame of ISAR image are located in the target region with higher energy,the ISAR image sequence accumulation energy optimization function of the triaxial stable targets and the fixed-axis slowly rotating targets are constructed,respectively.The 3-D imaging problem is transformed into an unconstrained optimization problem,which can be solved through the quantum-behaved particle swarm optimization(QPSO)algorithm iteratively.After that,the3-D coordinates of the scatterers and the rotational motion parameters of the fixed-axis slowly rotating target can be acquired.Finally,the effectiveness and robustness of the proposed method are verified by simulated data-based experiments.3.Aiming at the disadvantages that 3-D ISAR imaging methods based on image sequence energy accumulation is only suitable for space targets with a known motion priori,this dissertation extended the classic factorization framework and proposed a new 3-D imaging method for space targets.Firstly,the instance segmentation algorithms based on deep learning can be used to realize the extraction and association of key feature points throughout the multi-view ISAR images,and the range-Doppler(R-D)measurement matrix can be obtained.Then,the extended factorization framework(EFF)is proposed,which directly decomposes the unscaled R-D measurement matrix into the 3-D coordinates of the key feature points and the projection vectors from the 3-D space to multi-view ISAR images.Since the projection vectors are independent of scatterers,after the singular value decomposition,they can be used to construct the 3-D projection geometry.Finally,the 3-D imaging problem of the remaining scatterers can be transformed into an unconstrained optimization problem and can be solved iteratively,so as to achieve 3-D imaging of the whole target.Experimental results based on electromagnetic simulated data and measured data verify the effectiveness and robustness of the proposed method.