【作者】 刘承兰；

【导师】 沈荣骏； 黎湘；

【作者基本信息】 国防科学技术大学 ， 信息与通信工程， 2012， 博士

【摘要】 干涉逆合成孔径雷达（Interferometric Inverse Synthetic Aperture Radar，InISAR）可以实现对远距离运动目标的三维成像。相比于传统ISAR成像技术，InISAR可以获得目标的三维结构信息，且对目标姿态的变化不敏感，能够为目标识别提供更加全面、更加稳定的目标形状与结构信息，具有重要的理论意义和工程应用价值。本文深入研究了InISAR三维成像的回波建模、系统特性及成像方法等问题，研究成果对InISAR成像系统的设计、配置和成像处理具有重要的指导意义。第一章阐述了论文的研究背景及意义，回顾了InISAR成像系统的发展概况，综述了InISAR成像技术的研究现状，指出了InISAR成像的发展趋势，并介绍了论文的主要工作。第二章研究InISAR三维成像的回波建模和图像配准方法。建立了L型三天线InISAR成像系统回波模型，阐述了InISAR三维成像的基本原理。针对同一基线上两天线之间干涉图像失配的问题，深入研究了二维图像失配量及其对后续干涉处理的影响，并提出了基于相位校正的图像配准新方法。该方法从参考距离选取的角度出发，对辅助接收天线的回波信号进行相位校正，补偿了由干涉天线位置差异引起的两幅ISAR图像之间的失配量，实现了两幅图像间的精确配准。所提方法同时适用于多天线配置，与传统方法相比具有精度高、可推广性强的优点。提出了InISAR参考距离选择方式中“分别聚焦”和“统一聚焦”的概念。针对实际存在的测距误差问题，建立了InISAR参考距离测量误差模型，并对存在测距误差时分别聚焦和统一聚焦两种方法对应的InISAR成像性能进行了比较，发现分别聚焦方法在解决图像失配和相位模糊等问题方面更具有优势，从而为后续参考距离的选择提供了依据。第三章研究斜视背景下的InISAR三维成像问题。从理论上揭示了斜视对InISAR成像性能的影响：一方面，斜视附加相位与待求参量存在耦合，增加了目标方位向和高度向坐标值求解的难度。另一方面，斜视角的存在引起目标纵向坐标的估计误差和三维像的扭曲。针对上述两个难点问题，提出了基于非线性最小二乘-坐标变换联合估计目标散射点坐标的新方法。该方法首先利用非线性最小二乘解决斜视附加项与待求参量的耦合问题，而后通过坐标变换校正得到真实的目标纵向坐标值，有效解决了斜视情况下运动目标的InISAR三维成像问题。针对三维成像结果中的“散射点簇”现象，提出了ISAR像的“峰值检测”预处理方法，该方法可以有效剔除虚假散射点，得到的InISAR三维成像结果更加清晰且更接近于目标真实模型，从而显著提高了三维成像质量。第四章研究单天线InISAR三维成像问题。提出利用单部天线实现InISAR成像的设想：基于时间孔径和空间孔径的等效原理，以时间上的累积来获得空间上的天线孔径和等效基线。针对单天线InISAR系统无法准确测量“等效基线”的问题，提出基于坐标变换-方程联立的单天线InISAR三维成像新方法，实现了从不同时间段ISAR像的干涉相位中联立求解目标散射中心的位置信息。从理论上分析了所提单天线InISAR成像算法的适用条件和成像性能，并从工程实用的角度，指出了该算法存在的问题和难点，为进一步研究奠定了理论基础。第五章研究舰船InISAR三维成像问题。针对高海情下的舰船目标三维成像问题，建立了机载雷达对舰船的观测模型，在此基础上提出了一种新的舰船InISAR三维成像方法。所提方法更符合实际场景，利用三天线结构避免了方位向定标，且具有不需要舰船目标的先验位置信息等优点。针对舰船InISAR成像中的最优时间选择问题，提出了基于多策略融合估计的最优成像时间选择新方法。该方法首先根据雷达系统参数估计初始时间窗，而后基于信息融合的思想，综合利用图像对比度、图像熵、散射点个数以及多普勒展宽等ISAR图像聚焦质量的定量评价指标完成最优成像时段选择，避免了单个评价指标的局限性，有效提高了InISAR三维成像的质量。第六章研究InISAR三维成像中的角闪烁抑制问题。从矢量叠加的角度分析了InISAR成像中角闪烁的产生机理，揭示了合成矢量相位对应的等效散射中心位置与真实多散射点的位置关系，为角闪烁的判别和抑制提供了理论依据。为了提高成像分辨率，针对多频带雷达信号融合成像中的相干配准问题，提出了基于雷达测量数据相关的多频段雷达信号融合成像相干配准新方法。相对传统方法，该方法不仅降低了算法复杂度，消除了建模误差，还有效提高了相位参数估计的精度。为了利用天线阵列的空间分割能力，针对斜视背景下L型天线阵列三维成像存在的问题，提出了基于AMOD-PD-ERVP的L型阵列三维成像新方法，克服角闪烁影响，实现了真正意义上的三维分辨。第七章总结全文，并指出了需要进一步研究的问题。本文的研究成果丰富了非合作运动目标的高分辨雷达三维成像理论，对InISAR三维成像技术的发展具有较强的理论意义与应用价值。更多还原

【Abstract】 Interferometric inverse synthetic aperture radar (InISAR) is able to carry outthree-dimensional (3-D) image of far-field moving targets. Compared with traditionalISAR imaging, InISAR reflects3-D construction of the target which is not sensitive totarget orientation, and is capable of providing a more reliable description of targetfeatures, resulting in great advantage to target identification. Therefore, InISAR3-Dimaging has become a promising technique in the radar signal processing community.This dissertation investigates problems such as theory model, system characteristics andimaging methods, which can be applied to practical system design and configuration.Chapter1illustrates the background and significance of this research, and reviewsthe development of InISAR imaging system. Then the current signal processingtechniques in InISAR imaging are summarized, and the development trend of InISARimaging is analyzed, followed with the introduction of main content in this dissertation.Chapter2investigates the InISAR3-D imaging model and image registrationmethod. We develop an L-shape three-antenna InISAR imaging model and detail theimaging principles. With regard to the image mismatching problem in InISAR system, aquantitative analysis of the offset and its influence on the following interferometricprocessing was studied. From the view of reference range selection, a novel imageregistration method based on phase correction is proposed to solve the mismatchingproblem in InISAR imaging. The proposed method is suitable for both three-antennaand multi-antenna configuration. We introduce “respective focusing (RF)” and “uniformfocusing (UF)” for reference range selection process of InISAR. An InISAR referencerange error model is built and a comparison of interferometric imaging capability forboth reference range selection methods is performed. It is concluded that RF is a betterchoice in conquering problems such as image mismatching and phase ambiguity, whichcan be seen as a theoretical reference for practical reference range selection.Chapter3focuses on the squint-mode InISAR3-D imaging algorithm. Theinfluence of squint on InISAR imaging is theoretically investigated. First, coupling ofthe squint additive phase and the target azimuth/altitude coordinates to be solved maymake the solution more difficult. Second, the squint angle may lead to estimation errorof the vertical coordinates and distortion of the ultimate image. Aiming at the above twoproblems, we propose a new method combining nonlinear least square (NLS) andcoordinates transform (CT) to estimate the target coordinates, which overcomesdeficiencies of traditional methods, and effectively solves the squint-mode InISAR3-Dimaging problem. We also suggest a “peak detection” method for ISAR pretreatment,which eliminates the illusive scatterers and restrains the scatterer clusters problem,resulting in improvement of the3-D image quality. Chapter4studies the one-antenna InISAR3-D imaging method. We propose theassumption of InISAR imaging with only one antenna: to acquire the equivalentaperture and baseline based on the equivalent relationship between the time and thespace. Aiming at the difficulty of measuring equivalent baseline for one-antennaInISAR system, we propose a one-antenna InISAR3-D imaging method based oncoordinate transform and equation union, where the target scatterer position informationis estimated from the interferometric phases between ISAR images obtained fromdifferent measurement intervals. The applicability and performance of the proposedone-antenna InISAR imaging method is analyzed. At last, the practical difficulties of theproposed method are pointed out and can be seen as the direction of further study.Chapter5discusses the ship InISAR3-D imaging problem. For the sake of solvingthe3-D imaging problem of ship target in heavy sea-state, we establish a practicalairborne observation model using three antennas for ship targets, and propose a novelship InISAR3-D imaging method based on the model. Three-antenna configuration ismore suitable to the interferometric imaging of non-cooperative moving target and canavoid the cross-range scaling problem. Moreover, the prior knowledge of ship positionis not necessary in the proposed method. A multi-strategy-fusion based optimum timeselection algorithm for ship InISAR imaging is given. The proposed scheme estimatesthe initial time window from the radar system parameters, and then utilizes themeasurements of ISAR image focus, such as the image contrast, the image entropy, thescatterer number and the Doppler spread, to finish the InISAR imaging optimum timeselection. The proposed method avoids the deficiency of single measure, and improvesthe quality of InISAR3-D image.Chapter6concentrates on the glint restraint problem of InISAR imaging. We studythe mechanism of glint in virtue of vector addition, and show the location of synthesisscatterer relative to the multiple true scatterers, which provides a theoretical referencefor detecting and restraining the glint phenomenon. To increase the resolution, wepropose a new data-based coherent compensation method of multiband radar signalfusion imaging, compensating for lack of mutual coherence among different radars.Compared with the existing model-based coherent compensation method, the proposedmethod is simpler and the precision of phase parameters estimation is higher becausethe modeling error is eliminated. Moreover, it has better anti-noise capability. In orderto exploit the space separation capability of antenna array, we present anAMOD-PD-ERVP based L-shape antenna array3-D imaging method in the presence ofsquint. The proposed method overcomes the glint problem, and a3-D resolution isindeed realized.Chapter7summarizes this dissertation and discusses the future work.The research results achieved in this paper enrich the high-resolution radar3-Dimaging techniques of non-cooperative targets, and will be valuable for development of InISAR3-D imaging both in theory and in practice.更多还原