【作者】 杨科；

【导师】 廖桂生；

【作者基本信息】 西安电子科技大学 ， 信号与信息处理， 2014， 博士

【摘要】 旋转式合成孔径雷达(ROSAR)是一种新型的雷达成像模式，它既保留了传统合成孔径雷达(SAR)全天候、全天时、电磁穿透等优点，又具有重访周期短、全视域成像等卓越性能，广泛应用于自然灾难应急救援、地下资源勘探、公共场所无损安检、战场监视以及低空火力支援等领域。直升机等旋翼飞行器是ROSAR模式最为典型的应用平台，ROSAR天线固定在刚性支架上，并指向外围随机翼一起做匀速圆周运动，天线旋转一周即能完成周围场景一次探测。该技术巧妙地利用了机翼旋转特性，仅需要一个天线的旋转运动，就能形成方位向圆弧形合成孔径，从而实现周围场景二维成像，极大地提高了飞行器的飞行安全。但是面对低空飞行任务，周围场景通常非常复杂，而获取的二维图像将可能错误地反映潜在威胁物的空间信息，为此开展ROSAR三维成像研究显得尤为必要。本文以满足低空空域安全飞行为目的，针对ROSAR三维成像中的关键问题和技术难点，围绕973项目“复杂低空飞行的自主避险理论与方法研究”、国家自然科学基金“机载毫米波雷达旋转合成孔径成像处理方法研究”等项目的研究任务，从ROSAR三维成像模型和三维成像方法等方面展开研究。具体的工作和贡献如下：1.介绍了传统的二维ROSAR模型，为后续的三维成像理论奠定基础。针对现有的二维成像方法运算量大、大方位角散焦问题，提出了一种基于频谱重构的二维ROSAR成像方法。通过分析ROSAR回波信号形式，对其直接进行傅立叶变换，获取二维精确频谱表达式。由于精确频谱的形式过于复杂，导致后续处理无法进行，折中考虑频谱的精确性和可操作性，采用四阶频谱重构的方法，获取形式简洁、保留较多信息的高阶频谱，并基于该频谱提出ROSAR距离徙动算法(RMA)和调频变标算法(CSA)。仿真实验表明，在方位角不大于90度情况下，频谱重构方法都能实现高精度快速成像。2.建立了ROSAR干涉成像模型，并重点分析了非理想情况载机平台的轴偏移问题。通过ROSAR技术和干涉方法相结合，要求载机平台在不同高度上进行悬停，每悬停一次获取一组场景数据，对录取的数据进行成像和干涉处理，可实现周围场景全视域三维成像。针对平台运动非理想情况，讨论了轴偏移对方位带宽、图像位置和干涉相位的影响。轴偏移导致干涉相位中增加一项附加相位，通过推导和分析附加相位形式，构建了一个相位补偿函数，由于该函数需要场景高度这一未知信息，预先假设所有场景高度为零，然后通过构建相位加权因子来消除高度置零的影响。最后对轴偏移的影响进行仿真实验，并获取了仿真场景三维高程图。3.建立了旋转上升合成孔径雷达(SSAR)三维成像模型，并提出一种适用于该模型的三维成像算法。在现有的二维ROSAR技术的基础上，SSAR利用载机平台上升在高度向上形成第二个合成孔径，天线受水平旋转和匀速上升的共同作用，在空中形成一个圆柱形合成阵面，同时结合距离向发射的宽带信号，使得该模型具有三维成像能力。关于三维成像算法，为了降低成像处理难度，首先分析方位向采样和高度向上升对天线位置的影响，并构建位置偏移补偿函数，将“旋转上升”模式简化为沿高度向“一步一悬停”模式。然后分析了简化模式的相位历程，计算高度向中心频率偏移量，推导波数域三维匹配函数，构建波数域插值函数，进而实现简化模式的三维成像。最后对该模型进行了性能分析，并通过仿真实验验证成像算法的有效性。4.建立了前行ROSAR(FMROSAR)三维成像模型，并针对该模型提出了一种三维成像算法。在前行状态下，天线指向前方区域发射宽带信号，同时受到平台前行和机翼旋转的共同影响，在空中形成一个水平合成阵面，结合距离向高分辨能力以及二维阵面分辨能力，FMROSAR可实现三维前视成像。FMROSAR成像算法包括模式简化和简化模式三维成像两大步骤。首先通过方位向偏移补偿和顺轨向偏移补偿，将前行模式简化为沿顺轨向“一步一停”模式。针对“一步一停”模式，又细分为距离-方位成像和距离-顺轨成像两个子步骤。在后者的成像过程中，分析了平台运动对方位角变化的影响，并给出了顺轨向有效聚焦的约束条件。同时考虑顺轨向大斜视成像情况，提出了一种改进的距离多普勒算法(RDA)。最后分析了上述模型的性能，并验证了成像算法的有效性。更多还原

【Abstract】 As a new kind of imaging mode, Rotor Synthetic Aperture Radar (ROSAR) hasbeen used widely for natural disasters rescue, mineral resources exploration,nondestructive testing, battlefield surveillance and low-altitude fire support due to suchadvantages as all-time, all-weather, electromagnetic penetration, short revisiting timeand omnidirectional imaging. Helicopters and other rotorcrafts are the most typicalplatforms for ROSAR technology. The antenna is fixed at the end of a rigid arm, pointsaround the platform and travels along the circular motion with the constant velocityrotor. In this way, an observation of the surrounding scene is completed while theantenna rotates a full circle. Using the blade rotation characteristic, ROSAR technologycan generate a circular synthetic aperture in the azimuth direction with one antenna, andrealize the two-dimensional (2-D) imaging of surrounding scene for guaranteeingrotorcraft safety. However, the acquired2-D image may show spatial information of thepotential hazards incorrectly in a low-altitude and complex environment, and thus it isabsolutely necessary to carry out research on the three-dimensional (3-D) imaging ofROSAR.To make low-altitude aircraft secure, and to solve the key problems in3-D imagingof ROSAR, we expand the work from signal models and imaging methods of3-DROSAR. The relevant work is supported by the National Basic Research Program ofChina (973Program) under Grant2011CB707000and the National Natural ScienceFoundation of China under Grant61101242. The main work and contributions of thisdissertation are presented as follows:1. The model of2-D ROSAR is introduced, which lays the foundation for thesubsequent research on3-D imaging. To improve the weakness of time-consuming andazimuth defocus in the existing algorithms, a new2-D ROSAR algorithm based onspectral reconstruction is presented. The expression of ROSAR echo signal is analyzed,and then the precise2-D spectrum is obtained by using of the Fourier transform directly.Due to the complex form of2-D spectrum, it is difficult to realize the later imagingprocess. Therefore, considering the accuracy and operability, an approximate andinformative spectral is acquired by the method of fourth-order spectral reconstruction,and based on which, the Range Migration Algorithm (RMA) and the Chirp ScalingAlgorithm (CSA) of ROSAR is proposed. Finally, simulation results show the accuracyand rapidity of the proposed method with the azimuth angle of less than90degrees.2. An interferometric imaging model of ROSAR is established, and the problem of axial deviation of platform is analyzed emphatically under non-ideal circumstance.By means of ROSAR technique and interference method, a set of surrounding data arecollected in a certain “hover” position, while another set of data in a new “hover”position. After imaging and interferometric processing, we can get the omnidirectional3-D imaging of the surrounding scene. As to the non-ideal motion of the platform, wediscuss the influence of axial deviation on azimuth bandwidth, image position andinterferometric phase. In particular, the deviation leads to an additional term in theinterferometric phase. To correct this additional phase, we derive its expression,construct the relevant compensation function, and define a weighted factor as a result ofheight information inadequate and nulling in the compensation function. In thesimulation experiment, the influence of axial deviation is analyzed and the digitalelevation model (DEM) of3-D scene is generated.3. A model of spiral synthetic aperture radar (SSAR) is established and acorresponding algorithm is proposed. Based on the existing2-D ROSAR technology,SSAR utilizes the ascent of platform to form a second synthetic aperture, and then itachieves the ability of3-D imaging due to the formation of cylindrical array, which isthe result from the interaction of horizontal rotation and uniform ascent, and thewide-band signal transmitted in the range direction. In Chapter four,3-D imagingalgorithm is presented for the SSAR model. Firstly, to reduce the difficulty of imagingprocess, we analyze the impacts of azimuth sampling and height ascent on antennaposition, construct the compensation function for spatial offset, and simplify the “spiral”mode into the “hover-and-go” mode in the height direction. Secondly, through thedetailed analysis of phase history of simplified model, the spectrum shift in the heightdirection is calculated, the bulk compression function and interpolation kernel in thewavenumber domain is derived, and thus the3-D imaging is implemented. Finally, theperformance analysis of SSAR model is performed, and the effectiveness of theproposed algorithm is validated through simulation results.4. A model of forward-moving ROSAR (FMROSAR) is established, and a3-Dimaging algorithm is proposed for this model. In the forward-moving state, the antennatransmits the wide-band signal to the front area; meanwhile, it can form a plane arrayunder the influence of platform movement and blade rotation. Therefore, FMROSARcan realize3-D forward-looking imaging owing to the high-resolution in the rangedirection and the2-D resolution of the plane array. The imaging algorithm forFMROSAR includes model simplification and3-D imaging for the simplified model.First of all, with the azimuth and along-track offset compensation, the “forward-moving” mode is simplified as “stop-and-go” mode in the along-trackdirection. In “stop-and-go” mode, the algorithm can be further divided as therange-azimuth imaging and the range-along-track imaging. As for the latter, the azimuthangle variation and its impacts on the matched filter are discussed, and an improvedRange Doppler Algorithm (RMA) is proposed to focus the data with high squint anglein the along-track direction. Finally, the performance of FMROSAR is analyzed, and thevalidity of this algorithm is verified.更多还原