【作者】 孙超；

【导师】 宋祖勋；

【作者基本信息】 西北工业大学 ， 信息与通信工程， 2017， 博士

【摘要】 合成孔径雷达(SAR)和逆合成孔径雷达(ISAR)成像技术具有远距离探测、全天时和全天候工作的特点,有效提升了雷达信息感知和数据获取能力,在国防和民用众多领域具有广泛的应用。受奈奎斯特采样定理的约束,SAR/ISAR在提高成像分辨率时,面临信号采样率高、数据量大等问题,特别在多天线、多极化、多视角等多维度观测时,问题更为突出。对于现代多功能雷达和组网雷达,用于单个目标宽带成像的时间和资源是有限的,通常会导致频率和孔径数据不连续或缺损,从而影响成像分辨率。近年来提出的压缩感知理论(CS)为解决传统雷达成像面临的问题提供了一种新思路,在降低雷达数据采样率、弥补数据缺损和改善雷达成像质量等方面展现出巨大的潜力。本文以压缩感知理论及其在雷达成像中的应用为基础,围绕基于压缩感知的高分辨雷达成像方法展开研究,主要致力于解决压缩感知在对复杂运动目标ISAR成像、多通道和大角度SAR成像应用中存在的问题。主要贡献及创新性成果如下:1.针对传统距离-瞬时多普勒成像方法存在分辨率较低、对数据完整性要求高的局限性,提出一种基于压缩感知的机动目标距离-瞬时多普勒成像方法。该方法利用目标场景的空间稀疏性,在时频域对回波信号进行稀疏化表征,利用构建的瞬时字典获得相应的距离-瞬时多普勒图像,并通过迭代加权增强目标能量同时抑制噪声。仿真数据和实测数据处理结果表明,该方法有效提高了成像分辨能力,在低信噪比和数据缺损情况下具有更加稳健的成像性能。另外,将该方法推广到机动目标干涉ISAR三维成像,通过干涉通道联合稀疏化表征,有效提升了目标散射点高度信息的估计性能。2.针对网格误差和自旋角速度估计不精确带来的成像性能下降问题,首先提出了一种鲁棒的压缩感知窄带雷达成像方法,通过修正成像模型和改进优化重建算法,有效改善了网格误差对成像性能的影响;其次提出了一种联合自旋角速度估计的压缩感知成像方法,利用成像结果的图像熵对自旋角速度进行迭代搜索,实现了角速度的最佳估计和目标图像的最优重构。针对高速自旋目标常规宽带雷达三维成像方法存在计算复杂度高的问题,提出了一种新的自旋目标三维成像方法,利用窄带雷达成像方法和干涉技术实现目标三维成像,有效降低了成像复杂度。3.传统多通道SAR成像是对各通道回波数据进行独立成像处理,忽略了各通道间的相关性,不能保证多通道图像间散射点位置和数目的一致性。针对此问题,提出了一种基于联合稀疏重建的多通道SAR高分辨成像方法。该方法利用多通道图像具有相同稀疏性支撑这一先验信息,构建多通道信息联合约束函数,实现散射中心位置和数目在各通道下的一致性,并利用改进的优化重建算法进行多通道SAR成像联合优化重构。实验结果表明,该方法不仅能够利用少量的采样数据获得高分辨目标图像,而且保证目标散射点位置和数目在不同通道下的一致性,有效提高了目标信息的获取能力。4.大角度SAR传统子孔径综合成像方法容易受配准误差和目标各向异性的影响,导致综合图像中存在较多杂散点,针对此问题,提出了一种联合目标空间重构和方位依赖散射信息提取的子孔径图像联合稀疏重建方法。该方法利用子孔径图像间的相关性,定义综合全孔径信息的混合范数,并构建由混合范数约束的子孔径图像联合重建成像模型,实现了子孔径图像间的配准,有效抑制了综合图像中的杂散点;考虑到相邻子孔径间目标散射特性的缓变性,对生成的子孔径图像沿方位维进行平滑约束,使得提取的目标方位依赖散射信息更加准确可靠。更多还原

【Abstract】 Synthetic aperture radar(SAR)and inverse synthetic aperture radar(ISAR)imaging technique has the characteristics of long range detection,all-time and all-weather work,which effectively enhance the ability of information sensing and data acquisition of the modern radar.Therefore,SAR/ISAR imaging technique is widely used in many military and civil fields.As SAR and ISAR improve the imaging resolution,restricted by the Nyquist sampling theorem,they face the problems of high sampling rate and large amount of data.The problems become more prominent when it comes to multidimensional observations,such as multiple antennas,multiple polarizations and multiple views.For modern multi-function and networked radar,the time and resources used for wideband imaging of a single target are limited,resulting in the data discontinuity and missing,thereby influencing the imaging resolution.The newly proposed theory of compressed sensing(CS)provides a new way to solve the above problems in traditional radar imaging,which has shown tremendous potential in reducing the sampling rate of the radar data,making up for the data missing and improving the quality of image reconstruction,etc.Based on the CS theory and its application in radar imaging,this dissertation studies high-resolution radar imaging methods based on CS to solve the problems of the CS applications in ISAR imaging of complex targets,multichannel SAR and wide angle SAR imaging.Main contributions and innovative achievements of this dissertation are as follows:1.As traditional range-instantaneous Doppler(RID)imaging method is of low resolution and has high requirements for the integrity of the data,a range-instantaneous Doppler(RID)imaging method of maneuvering targets based on CS is proposed.Exploiting the spatial sparsity of the target scene,the target echo is sparsely represented in the time-frequency domain;the instaneous dictionary is constructed to obtain the range-instaneous Doppler images of different moments;in order to enhance the energy of the real target and suppress the noise,the strategy of weighted iteration is introduced.Both simulation and experimental results show that this method has higher resolution and more robust imaging performance in the case of low signal-to-noise ratio and the data missing.Furthermore,this method is extended to interferometric ISAR three-dimensional(3D)imaging of maneuvering targets.The interferometric channels are jointly sparsely represented to enhance the estimation performance of the height information of scatterers.2.Due to the grid error and inaccurate estimation of the spinning angular velocity,the imaging performance is affected considerably.In order to solve this problem,a robust narrowband radar imaging method based on CS is proposed firstly,which can effectively alleviate the influence of the grid error on imaging performance,by optimizing the imaging model and improving the reconstruction algorithm.Secondly,a joint method with CS imaging and the optimal estimation of spinning angular velocity is proposed,which can obtain the best estimation of the spinning angular velocity and the optimal reconstruction of target,by iterative research of the spinning angular velocity based on the image entropy of the imaging result.In order to solve the problem of high computational complexity of conventional wideband radar 3D imaging methods of spinning targets,a new 3D imaging method of spinning targets is proposed.This method combines narrowband radar 2D imaging and interferometric processing to obtain 3D images of the target,which effectively reduce the complexity of the imaging method.3.Traditional multichannel SAR imaging uses independent processing of each channel,which ignores the correlation between channels and cannot guarantee the consistency of the positions and the number of scatterers in multichannel images.To solve this problem,a multichannel SAR high-resolution imaging method based on joint sparse reconstruction is proposed.This method uses the prior information of the same sparsity support over multichannel images,constructs the joint constraint function integrating multichannel information to realize the consistency of the positions and the number of scatterers in different channels,and uses the improved reconstruction algorithm to achieve the joint reconstruction of multichannel SAR images.The experimental results show that this method can obtain high-resolution SAR images with limited measurements and the positions and the number of scatterers are aligned in different channels,which improves the ability of information acquisition of the target.4.Traditional subaperture composite imaging method of wide-angle SAR is easily affected by the registration errors and the anisotropic scattering behavior of the target,resulting in some desultory points existed in the composite image.To solve this problem,a joint sparse reconstruction method of subaperture images for joint target space reconstruction and azimuth dependent scattering feature extraction is put forward.This method exploits the correlation between different subaperture images,defines the mixed norm with comprehensive information of full aperture,and constructs the joint sparse reconstruction model of all subaperture images constricted by the mixed norm,which can realize the precise match and suppress desultory points in the composite image.Considering the continuity of the target scattering characteristics between the adjacent subapertures,a smoothness constraint for subaperture images along the azimuth dimension is used to make the extracted azimuth dependent scattering characteristics more accurate.更多还原