【作者】 刘燕；

【导师】 邢孟道；

【作者基本信息】 西安电子科技大学 ， 信号与信息处理， 2012， 博士

【摘要】 合成孔径雷达（Synthetic Aperture Radar，SAR）成像技术具有全天候、全天时和远距离成像的特点，有效地提高了雷达的信息获取能力，具有重大的军用和民用价值。获得高质量和高分辨率图像是SAR成像追求的目标。相对于低分辨率SAR而言，高分辨率SAR要求成像处理更为精确，因此需要构建精度更高的成像模型。本文围绕高速平台高分辨率SAR成像、大场景超高分辨星载SAR成像、圆轨迹SAR成像、宽测绘带高分辨率动目标成像和运动参数估计这几个方面作了一些工作，具体如下：1.高速平台高分辨率SAR回波信号分析以及成像处理目前SAR成像处理大多采用了对雷达载体运动的“一步一停”假设。“一步一停”假设忽略了两种雷达运动，一种是发射和接收期间的雷达运动，另一种是信号传输期间的雷达运动。忽略这两种雷达运动会使成像处理存在一定误差，随着SAR分辨率的提高，这些误差有可能会影响成像效果。针对这一问题，推导了“一步一停”假设给成像带来的误差，包括：距离向二次相位误差、距离徙动误差、方位向三次相位误差以及方位向偏移，同时，给出了“一步一停”假设的适用范围。分析得到：当雷达载体为高速运动平台（例如，卫星或高空超音速飞行器）时，在分辨率很高的情况下，“一步一停”假设带来的误差会影响成像质量。在一个方位向分辨率为0.2m的星载聚束SAR仿真实验中，展示了“一步一停”假设给成像带来的不可忽略的误差，仿真实验验证了分析的有效性。为了保证大测绘带宽以及降低系统数据传输的压力，脉冲重复频率PRF不能过高，但高分辨率成像又需要大方位带宽，因此接收到的信号通常是方位频谱混叠的。为了利用条带SAR成像处理方法对回波进行精确成像处理，需要得到方位频谱不混叠的数据。所提出的高速平台高分辨率聚束SAR的成像算法借鉴了“两步”算法的思想，“第一步”采用随距离频域空变的方位deramp函数进行方位deramp处理来去除方位频谱混叠；“第二步”结合所提出的精确回波模型的频域表达式，采用ω-k算法对方位频谱不混叠的数据进行成像处理。方位预处理仿真实验展示了随距离频率空变的方位deramp函数在方位解模糊时的优势。点目标SAR成像仿真实验验证了所提成像方法的有效性。2.基于修正ω-k算法的大场景超高分辨星载SAR成像处理方法在考虑了星载SAR等效斜视模型误差补偿以及“一步一停”假设带来误差的补偿的情况下，采用星载等效斜视模型推导了一种基于改进ω-k算法的大场景高分辨率SAR成像算法。成像算法针对传统ω-k算法中的Stolt插值无法在距离频域处理等效速度和等效斜视角沿距离向的变化这一问题，首先，分析了采用恒定的等效速度进行Stolt插值给距离徙动校正带来的误差，并给出一个超高分辨率下该误差随场景距离向宽度变化的实例，指出对传统Stolt插值进行修正的必要性；其次，通过分析等效速度以及等效速度的平方随距离的变化规律，对传统Stolt插值进行了修正；最后，基于修正的Stolt插值，推导得到了沿距离向的成像指标一致的修正ω-k成像算法。仿真实验验证了所提算法的有效性。3.圆轨迹SAR快速成像由于圆轨迹SAR特殊的运动轨迹，直线SAR的成像算法不能直接应用于圆轨迹SAR数据处理。然而，对于圆轨迹SAR系统而言，其响应函数具有“沿角度维平移不变”的特性。因此可以利用这一特性，借鉴条带SAR成像算法的思想，在频域研究圆轨迹SAR数据的快速成像方法。本文推导了完全精确的柱面坐标系下圆轨迹聚束SAR和圆轨迹环视SAR的回波频谱解析表达式。利用条带SAR成像算法的思想，通过对频谱的合理近似，给出了圆轨迹SAR的成像算法。仿真实验验证了算法的有效性。4.低重频采样SAR系统中地面运动目标运动参数估计与成像基于压缩感知理论，本文提出一种低重频采样SAR系统中地面运动目标高分辨率成像以及运动参数估计的新方法。该方法利用距离脉压后的数据进行杂波相消以此保留了运动目标信息；去除由雷达载体速度产生的多普勒调频项后，采用基于keystone变换的稀疏走动校正的方法对多个运动目标的包络进行走动校正；最后，基于压缩感知理论，本文构造冗余基矩阵，将空间稀疏分布的目标运动参数估计以及成像问题转化为求优化方程的稀疏解，从而实现运动目标的准确重建。更多还原

【Abstract】 Synthetic Aperture Radar (SAR) imaging is a powerful and well established microwaveremote sensing technique which can be used for long distance imaging and enables highresolution measurements of the earth surface independent of weather conditions and sunlightillumination. It is producing high resolution and high quality images that SAR imagingpursuits. Compared with low resolution SAR, high resolution SAR data needs more acurrateimaging formation model and more precise imaging processing. This dissertation is devotedto imaging for High Resolution SAR on High-Speed Platform, imaging for wide field andhigh resolution spaceborne spotlight SAR, fast imaging for circular SAR, imaging and motionparameters estimation for moving targets in high resolution SAR system with wide swath. Themain research results are as follows:1. Echo Model Analyses and Imaging Algorithm for High Resolution SAR on HighSpeed Platform“Stop-go-stop” approximation of radar carrier simplies the processing of SAR imaging,which is widely used in the SAR systems. There exist two kinds of neglected motion in thegeometric configuration of SAR based on the “stop-go-stop” approximation: the motionoccurring between transmission and reception of a pulse and during the interval transmissionand reception of a pulse, so that the imaging processing based on the “stop-go-stop”approximation is not accurate. With the increasing requirements on high resolution imaging,the errors brought by the “stop-go-stop” approximation may be intolerable for SAR imaging.The errors caused by the “stop-go-stop” approximation are introduced, including the quadraticphase error in range, the range cell migration error, the cubic phase error in azimuth and thelocation deviation of the targets, and then the application scope for the “stop-go-stop”approximation is presented correspondingly. A reference system of the spaceborne spotlightSAR with0.2m azimuth resolution is given for simulation to illustrate the intolerable errors.In order to make sure the coverage performance of the system and reduce the pressure todata transmission system, the pulse repetition frequency (PRF) should not be too high, whichmakes it difficult to exceed the Doppler bandwidth and will induce azimuth spectrum aliasing.In order to use the precise imaging processing of the stripmap SAR, the azimuth spectrum ofthe data should not be aliasing. Based on the accurate echo model, a spotlight imagingalgorithm for the high resolution SAR on high-speed platform is presented. The proposedalgorithm takes advantage of the “two-step SAR focusing strategy” and the rangefrequency-dependent reference function is adopted in the azimuth preprocessing. The simulations for the azimuth preprocessing illustrate the advantage of using the rangefrequency-dependent reference function in the azimuth preprocessing for high resolution SARon the high-speed platform. Point target simulations show the performance of the proposedimaging algorithm.2. A modifiedω-k algorithm for wide field and high resolution spaceborne SAR imagingAfter considering the compensation of the errors brought by equivalent-squint rangemodel and the errors brought by the “stop-go-stop” approximation, this dissertation producesa modifiedω-k algorithm for wide field and high resolution spaceborne spotlight SAR. Thevariation of the equivalent-velocity and the equivalent-squint angle along the range directioncannot be taken account into the Stolt mapping of the traditionalω-k algorithm. Before theproposed algorithm is derived, the error of the range migration correction (RMC) by using theuniform equivalent-velocity in the Stolt mapping is analyzed, and the necessity for modifyingthe traditional Stolt mapping is presented. In order to process the wide field and highresolution spaceborne SAR data precisely, by utilizing the relationship between the range andthe equivalent velocity, the Stolt mapping of the classical ω-k algorithm is modified. Themodified Stolt mapping takes the range-dependence of the equivalent velocity into account, sothat the range cell migration is corrected accurately. Using the modified Stolt mapping, themodifiedω-k algorithm is presented, which guarantees the uniform image quality along rangeand is validated by the simulation.3. Fast imaging processing of circular SARBecause of the special track of circular SAR, the imaging algorithms of SAR in lineartrack cannot be used in processing the circular SAR data directly. However, the responsefunctions of the circular SAR systems can remain the consistency by changing the sameaspect angle both of the target and the starting position of the radar. Because of thischaracteristic, circular SAR can be processed in the azimuth frequency domain by using thefast imaging algorithm of SAR in linear track for reference. In this dissertation, the exactsignal spectrum expressions of circular spotlight SAR and circular looking around SAR arederived. By using the reasonable approximations of the exact signal spectrum expressions, thefast imaging algorithms are presented. The simulations show the good performance of theproposed algorithms.4. Motion parameters estimation and imaging for moving targets in SAR system with alow PRF sampling rateBased on the CS theory, a novel method is proposed for the high resolution imaging andthe motion parameters estimating of the ground moving targets in the SAR system with lowPRF sampling rate. Firstly, the clutter cancellation is used to acquire the information of the moving targets in the range compression data domain. Secondly, after the azimuth dechirpprocessing, the range-walk of each target envelope is corrected by the Keystone transform andthe sparse range-walk correction filtering. Thirdly, the redundant base matrix is constructedaccording to the azimuth signal expression, and then, based on the CS theory, the motionparameters estimation and imaging for the sparse targets can be resolved by finding thesparsest solution of the optimization equation. The simulations show the performance of theproposed algorithm.更多还原