【作者】 郑晓双；

【导师】 禹卫东；

【作者基本信息】 中国科学院研究生院（电子学研究所） ， 通信与信息系统， 2007， 博士

【摘要】 运动补偿是高分辨率机载合成孔径雷达(Synthetic Aperture Radar，SAR)成像处理中不可或缺的一部分，已成为机载SAR领域的一个重要研究内容。论文介绍了机载SAR运动补偿技术的基本原理和实现方法，对基于数据处理的运动补偿技术进行了深入研究。基于数据处理的运动补偿技术分为两类：已知运动误差的运动补偿和未知运动误差的运动补偿。本文对这两类技术分别进行了研究。另外，论文还开展了用现场可编程门阵列(FPGA)实现实时运动补偿及成像处理的方案设计和硬件设计工作。论文主要内容包括五部分：(1)研究了五种典型的SAR成像算法，包括适于条带SAR成像的距离-多普勒算法(R-DA)、Chirp Scaling算法(CSA)和距离徙动算法(RMA)，以及适于聚束SAR成像的极坐标格式算法和基于极坐标格式的重叠子孔径算法。对这几种算法的原理、实现流程作了详细介绍，并根据距离徙动量的大小对不同算法的适用范围和成像性能进行比较。(2)对机载SAR运动误差的不同类型进行了详细分析，重点探讨了天线相位中心位置偏差引起的回波信号相位误差。定量分析了不同形式的相位误差对SAR图像质量的影响，并给出了相应的处理方法。最后对机载SAR的运动补偿技术进行分类。(3)针对运动误差二维空变性的问题作了全面讨论，并对已知运动误差的运动补偿中包络校正、一次运动补偿、二次运动补偿、窄波束运动补偿和宽波束运动补偿的不同要求和补偿方式给予界定。研究了三种窄波束运动补偿算法：结合R-DA的运动补偿算法、结合CSA的运动补偿算法和结合RMA的运动补偿算法。同时对宽波束运动补偿技术进行深入研究和改进，提出了三种条带模式的宽波束运动补偿算法：频域逐块补偿算法、重叠保留分块运动补偿算法、基于频域分割的运动补偿算法和一种聚束模式的极坐标格式重叠子孔径运动补偿算法。其中频域逐块补偿算法采用频域补偿，对低频误差具有良好的补偿效果，同时取得了运算量和补偿精度的折中；后三种算法采用时域补偿的方式，能有效的补偿二维空变的低频和高频运动误差的影响。(4)针对未知运动误差的运动补偿方法，主要研究了自聚焦。自聚焦方法包括三类：子孔径法、逆滤波法和基于代价函数的方法。文中重点介绍了逆滤波方法中的相位梯度自聚焦算法和基于代价函数方法中的最小熵算法。(5)采用Xilinx的Virtex-Ⅱ系列FPGA设计并实现了星载SAR星上实时成像处理器，给出了点目标实测结果；设计了基于FPGA的机载SAR实时运动补偿和成像系统，应用单片Virtex-Ⅳ芯片实现，完成了128×128点数据的成像仿真测试。更多还原

【Abstract】 Motion compensation (MOCO) is an essential part of high-resolution airbornesynthetic aperture radar (SAR) imaging. MOCO is an important research field ofairborne SAR signal processing. This thesis introduces the rationales and theimplementation of airborne SAR MOCO, and then studies the MOCO based on dataprocessing. MOCO based on data processing includes two categories: the one isMOCO for known motion errors and the other is that for unknown ones. This thesismakes some discusses about these two MOCO. In addition, it finishes the softwareand hardware design of real-time MOCO and imaging based on field programmablegate arrays (FPGAs). The thesis is comprised of five main components as follows:(1) Five typical image formation algorithms are studied, which are three algorithmsfor stripmap SAR: range-Doppler algorithm (R-DA), chirp scaling algorithm(CSA), range migration algorithm (RMA), and two for spotlight SAR: polarformat algorithm (PFA) and polar format based overlapped subaperture algorithm(PF-OSA). The rationales and processing steps of these algorithms are introducedin details. Comparison of their applicabilities and performances is made accordingto the quantities of range cell migration.(2) Different types of airborne SAR motion errors are analyzed, which are caused bythe deviations of the antenna phase centre. Since different phase errors havedifferent impacts on the SAR images, corresponding solutions are proposed.Finally the techniques of MOCO are classified and summarized.(3) Two-dimensional space-variant motion errors are studied comprehensively.Envelope correction, first-order compensation, second-order compensation,narrow beam compensation and wide beam compensation are discussed individually. Three narrow beam MOCO algorithms are deeply studied, includingMOCO integrated in R-DA, in CSA and in RMA. Some innovative research ismade in the wide beam MOCO. Three different wide beam MOCO algorithms areproposed for stripmap SAR, which are block by block compensation in frequencydomain, overlapped block processing and frequency division based MOCO.PF-OSA integrated with MOCO is an algorithm for spotlight SAR. The block byblock compensation achieves computation efficiency and precisioncompromisingly in the case of low-frequency errors. The last three ones adoptcompensation in time domain and can correct two-dimensional space-variant lowand high-frequency motion errors.(4) Auto-focus is an important motion compensation method of MOCO for unknownerrors. It consists of three kinds: sub aperture processing, inverse filtering andcost based approaches. Phase gradient auto-focus and entropy based auto-focus(stage by stage approaching, SSA) are introduced with simulation.(5) Real-time MOCO and imaging based on FPGAs are considered at last. On-boardreal-time imaging processor for space borne SAP, is implemented with Virtex-Ⅱchips from Xilinx Corporation. Test result of a point target is shown. Meanwhile,real-time imaging processor with MOCO for airborne SAR is also designed. Thewhole design is implemented on a single chip Virtex-Ⅳand simulation of a blockof data (128×128) is finished.更多还原