【作者】 王昕；

【导师】 朱岱寅；

【作者基本信息】 南京航空航天大学 ， 通信与信息系统， 2011， 博士

【摘要】 合成孔径雷达(SyntheticAperture Radar,简称SAR)是一种全天时、全天候的高分辨率微波成像系统。经过几十年的发展，常规分辨率的单基SAR成像技术已经逐渐趋于成熟。但是，应用现有成像技术聚焦X波段超高分辨率SAR或者双基SAR回波数据时仍然存在很多问题。超高分辨率条件下，线性调频变尺度算法(Chirp Scaling Algorithm,简称CSA)等近似频率域算法的相位误差变大，导致SAR图像聚焦质量显著下降。针对此问题，本文研究了在步进调频体制下对两种近似频率域算法进行改进，减小算法相位误差，从而实现X波段超高分辨率SAR回波数据的高精度聚焦处理。双基SAR系统中，发射机和接收机分置使得回波信号的多普勒相位历史较为复杂，高精度的双基成像算法的推导也较为繁琐。本文对双基极坐标格式算法(Polar Format Algorithm,简称PFA)和双基反投影算法(Back Projection Algorithm,简称BPA)进行了分析和完善。此外，对现有几种近似双基频谱的精度进行了比较，并推导了双基距离多普勒算法(Range DopplerAlgorithm，简称RDA)。论文第一章绪论，介绍了本文工作的研究背景，回顾了单基和双基SAR的发展历程和现状。最后，总结了本文的主要工作。论文第二章研究了基于步进调频信号的超高分辨率SAR成像算法。本章首先介绍了步进调频回波信号模型以及现有带宽合成方法。然后，分别讨论了直接接收和dechirp接收方式下回波信号的处理流程。该流程利用CSA或者扩展频率域变尺度算法(Extended Frequency ScalingAlgorithm,简称EFSA)聚焦接收子脉冲信号集，在二维频率域对子脉冲集进行拼接，并对拼接所得数据做二维逆傅立叶变换得到图像。最后，对上述处理流程和原始CSA/EFSA聚焦全带宽信号的相位误差进行了分析和比较。点目标仿真和相位误差分析均表明，本章提出的子脉冲数据聚焦后拼接的处理流程可以有效的改善图像聚焦质量。论文第三章讨论了基于尺度变换原理的双基PFA。首先简单介绍了双基聚束模式下的成像几何关系以及双基PFA流程。然后，讨论了尺度变换原理(Principle of Chirp Scaling，简称PCS)以及双基PFA距离向重采样的本质。最后，对直接接收和dechirp接收方式下的回波信号分别进行了距离向变标处理。由于应用尺度变换代替插值来实现数据的距离向重采样，双基PFA的运算效率得到了有效改善。论文第四章对双基PFA的波前弯曲误差进行了分析和校正。平面波前假设在双基PFA聚焦过程中引入了相位误差，导致图像出现空变几何失真和散焦现象。本章首先推导了波前弯曲误差的解析表达式。然后，基于此解析表达式，设计空变滤波器来补偿双基PFA图像中的散焦现象，并利用插值运算来校正图像中的几何失真。点目标仿真以及滤波后场景聚焦范围的分析表明该方法可以有效地补偿双基PFA的波前弯曲误差。论文第五章讨论了双基滤波反投影(Filter Back Projection，简称FBP)算法以及FBP算法与CSA之间的联系。基于电磁场模型和Born近似，SAR基带回波信号被视为场景复反射系数的傅立叶积分变换，对信号应用FBP类的逆转方法可以得到图像。本章进一步完善了该FBP方法在双基SAR成像中的应用，并对其本质进行了分析。然后，在单基配置下，对FBP算法与CSA之间的联系进行了推导。发现在忽略信号距离向调频斜率的空变性的条件下，FBP算法可以近似得到CSA处理流程。这也进一步验证了FBP的二维空变匹配滤波器本质。论文第六章对双基二维点目标响应频谱和双基RDA进行了分析。首先，简单介绍了现有三种近似的二维点目标响应频谱，并对频谱精度进行了分析和比较。分析表明改进后的Loffeld双基公式(Improved Loffeld’s Bistatic Formula，简称ILBF)与级数逆转(Method of SeriesReversion，简称MSR)频谱精度相当。基于ILBF频谱，本章推导并分析了双基RDA。与现有双基频率域算法相比，该算法形式上简洁，并可以适用于双基方位向空变以及空不变配置下。最后，仿真实验验证了该算法的有效性。更多还原

【Abstract】 Synthetic aperture radar (SAR) is a high resolution microwave imaging system with all weather,day and night capability. After the developments in the past decades, the monostatic SAR imagingtechnologies in general resolution case become mature gradually. However, there are still someproblems when utilizing the existing algorithms to focus the X band ultra-high resolution SAR data orbistatic SAR data.In ultra-high resolution case, the phase errors in the approximated frequency domain algorithms,such as chirp scaling algorithm (CSA), become larger, which degrade the focal quality of SAR imagesignificantly. To solve this problem, this dissertation investigates the modification of two frequencydomain algorithms in ultra-high resolution SAR imaging with the transmitted stepped frequencywaveform. With this modification, the phase errors in these algorithms are reduced so as to focus theX band ultra-high resolution SAR data with high precision.In bistatic SAR, the separation of transmitter and receiver results in a complicate form of theDoppler phase term of the received signal, and hence the derivation of bistatic algorithm with highprecision is complex. The bistatic polar format algorithm (PFA) and bistatic back projection algorithm(BPA), which are obtained by extending the corresponding monostatic algorithms, are analyzed andimproved in this dissertation. Then, the existing approximate bistatic spectrums are analyzed andcompared, and the bistatic range Doppler algorithm (RDA) is derived.Chapter I is the introduction, which introduces the background of this paper and describes thehistory and latest developments in monostatic and bistatic SAR. Then, the difficulties and maincontents in our research are outlined.Chapter II investigates the ultra-high resolution SAR imaging with the transmitted steppedfrequency waveform. This chapter first introduces the signal model of received stepped frequencywaveform and existing bandwidth combination methods. The new processing flows for the receivedchirp and dechirped stepped frequency signal are then proposed and discussed, respectively. The CSAand extended frequency scaling algorithm (EFSA) are applied to focus the sub pulses, respectively.After combining the focused sub pulses in the2-D frequency domain, the image can be obtained viaperforming2-D IFFT on the combined data. The phase error in the above processing flows and in theoriginal CSA/EFSA are analyzed and compared finally. Point target simulation and phase erroranalyses indicate that our proposed methodology can improve the focal quality of image effectively. In Chapter III, the implementation of bistatic PFA using the principle of chip scaling (PCS) isproposed. The bistatic spotlight mode SAR data collection geometry and bistatic PFA are introducedfirst. Then, the PCS and essence of range resampling in bistatic PFA are discussed. Finally, the scalingprocessing flows for the chirp and dechirped signal are described, respectively. Since the scalingmethodology is used to implement the range resampling, the efficiency of bistatic PFA is improvedeffectively.Chapter IV analyzes and corrects for the wavefront curvature error in bistatic PFA image. Theunrealistic planar wavefront assumption in bistatic PFA introduces phase error in the focused imageand results in space variant geometry distortion and defocusing effect. This chapter first derives theanalytical expression of wavefront curvature. Then, the space variant filter is designed and applied tocompensate for the defocusing effect in bistatic PFA image, and interpolation operation is used tocorrect for the geometry distortion. Analysis of the focused scene size after space-variant filteringvalidates this algorithm.Chapter V studies the bistatic filtered back projection (FBP) algorithm and the link betweenFBP and CSA. Based on the microwave model and Born approximation, the received baseband signalin SAR can be considered as the fourier integral transform of the reflectivity of the illuminated scene.The image can be obtained by applying a FBP-type inversion algorithm. This chapter completes theapplication of FBP algorithm in bistatic SAR imaging, where the essence of this methodology is alsodiscussed. Moreover, the link between FBP and CSA in monostatic case is derived. When ignoring therange variance of range modulation rate, the processing flow of FBP can be approximated as that ofCSA, which also indicates that the essence of FBP is a2dimensional space variant matched filter.In Chapter IV, bistatic point target response spectrum and bistatic RDA are analyzed. Threeexisting approximate spectrums are analyzed and compared first. The improved Loffeld’s bistaticformula (ILBF) spectrum is proved to be comparably accurate with the spectrum derived using themethod of series reversion (MSR). Based on the expansion of the ILBF spectrum, a new bistatic RDAis developed to process the azimuth invariant and variant bistatic SAR data. Compared with theexisting bistatic frequency domain algorithm, the new algorithm has a simpler formulation, and is ableto cope with bistatic SAR data in azimuth invariant and variant configurations. Finally, point targetssimulations validate the new algorithm.更多还原