Research On Downward-looking Linear Array Three-Dimensional SAR Imaging Technology

Synthetic aperture radar imaging has been nowadays used in a wide range of civil and military applications,due to the advantages like all-weather capabilities and independence of daylight.Since the conventional SAR works in the side-looking mode,it is inevitable to result in the shading and layover.To overcome these restrictions and acquire the whole information of the scene,SAR has worked in the downward-looking mode.However,downward-looking SAR usually produces the blind spot beneath the platform and the left/right ambiguity in the cross-track direction.Fortunately,it can be solved by beamforming operation with a linear array distributed along cross-track direction.Therefore,downward-looking linear array three-dimensional SAR(DLLA 3-D SAR)is presented to not only overcome the restrictions of the conventional SAR,but also map a directly overflown scene into a high resolution three-dimensional imagery and acquire completer and more detailed information of the targets.Aiming at improving the imaging resolution,this dissertation studies the theory and techniques needed in downward-looking linear array three-dimensional SAR imaging to solve the problems existed in the imaging processing.Chapter 1 illustrates the research background and significance of this dissertation,briefly recalls the development of downward-looking linear array three-dimensional SAR systems,summarizes the existing imaging methods and the related works,and generalizes the development on the super-resolution imaging techniques and its applications on downward-looking linear array three-dimensional SAR.Finally,the main research work and organization of this dissertation is outlined.Chapter 2 describes the basic principle and theory of downward-looking linear array three-dimensional SAR imaging and three-dimensional imaging algorithms.Firstly,it derives the signal model of downward-looking linear array three-dimensional SAR imaging;then the conventional RD,?-k and BP algorithm are expanded into the three-dimension.Three-dimensional imaging algorithms such as 3-D RD,3-D ?-k,and 3-D BP imaging algorithms have been proposed and compared.Subsequently,it points out some key techniques existed in three-dimensional SAR imaging,which provide the theoretical basis and research orientation for the following researches.Chapter 3 studies downward-looking linear array three-dimensional SAR imaging based on the uniform linear array,which is conducted to solve the problem of much lower resolution in the cross-track direction.Firstly,the matrix format of signal model for downward-looking linear array three-dimensional SAR is derived,which can be reformulated the focusing in the cross-track as a spatial spectrum estimation issue.Hence,the super-resolution imaging algorithm based on FFT-MUSIC is presented to estimate the positions of the scattering centers and acquire the imagery of targets.By the nearby spatial smoothing method,the responses from each scattering center can be decorrelated without the reduction of the effective aperture.Moreover,the FFT preprocessing can not only reduce the range of peak searching in MUSIC algorithm to improve the computation efficiency,but also directly obtain the backscattering coefficient.In addition,the effects caused by vibrations of the wings are discussed with respect to the 3-D imaging quality and can be conquered by a consistent motion compensation method.Finally,experimental results by simulation data and real data are shown that the proposed imaging algorithm can improve the resolution in cross-track direction without the effect of side lobes.Chapter 4 studies downward-looking linear array three-dimensional SAR imaging based on the non-uniform linear array.Firstly,it briefly recalls the basic theory,reconstruction conditions,and algorithms of compressive sensing(CS);then based on MIMO sparse linear array and T/R non-uniform sparse linear array,the signal models of downwardlooking linear array three-dimensional SAR imaging are derived.Combined with CS theory,the super-resolution imaging algorithm based on TSVD-CS has been proposed.The most attractive idea of CS is that it is possible to recover a signal sampled at a much lower rate than the Nyquist rate.It is not only capable of providing a super-resolution obtainable with the given linear array size,but also applicable to the non-uniform linear array.Moreover,in order to advance the efficiency of the imaging algorithm,TSVD method is used to alleviate the computational cost of the conventional CS by reconstruct the measurement matrix.Finally,results from simulation data shows that the proposed algorithm can achieve a significant improvement in computational efficiency and maintenance in accuracy and robustness of target separation with whatever the uniform linear array or the non-uniform linear array.Chapter 5 proposes downward-looking linear array three-dimensional SAR superresolution imaging via two-dimensional joint processing.The length limitations of synthetic aperture and linear array lead to the poor azimuth and cross-track resolution.Thus,the two-dimensional joint super-resolution imaging method is proposed,which can deal with the couple effect between different directions availably.Using joint sparse property,the imaging in azimuth and cross-track direction is cast as a two-dimensional sparse reconstruction problem.Then the super-resolution imaging algorithm based on two-dimensional compressive sensing is proposed.Compared with the conventional method by transforming two-dimensional matrix into a one-dimensional vector,the 2-D SL0 algorithm is presented to be used directly for sparse reconstruction of 2-D signals on dictionaries with separable atoms.This algorithm can reduce the computational cost and memory consumption.To further improve the computational efficiency,the two-dimensional joint sparse sampling strategy is defined to alleviate the computational cost.Since the 2-D undersampled data renders the Nyquist sampling theory invalid,the imaging results based on the conventional matched filter method is affected by aliasing and high-level side-lobes.Therefore,a novel DLLA 3-D SAR imaging algorithm based on matrix completion is proposed for under-sampled azimuth-cross-track data.The proposed algorithm can effectively suppress the adverse effects,including high-level side-lobes,even the appearance of the aliasing and false-alarm targets,caused by the under-sampled data.Finally,numerical simulations are presented to evaluate the limits of the proposed two algorithms under the noise scenarios.The simulation results show that the proposed algorithms can provide the high-resolution images with the less memory consumption and higher computational efficiency.Additionally,the results of the experiment on real data demonstrate the validity and the advantages of the proposed imaging algorithms.Chapter 6 concludes the research work and main innovations in the whole dissertation,and presents the outlook of the future work.