【作者】 张永红；

【导师】 林宗坚； 张继贤；

【作者基本信息】 武汉大学 ， 摄影测量与遥感， 2001， 博士

【摘要】 合成孔径雷达作为二十世纪出现的尖端对地观测技术,由于它具有全天时、全天候的成像能力并能穿透一些地物,在土地覆盖制图、生态和农业、固体地球科学、水文、海冰等众多领域有着广泛的应用。随着未来更高分辨率、多极化、多波段、更优化的干涉测量设计的SAR系统的出现,合成孔径雷达遥感技术将会在更多的领域扮演更重要的角色。合成孔径雷达遥感技术在我国有着极大的潜在应用市场,对于某些特殊问题的解决,例如西部困难地区的地形图测绘及南方阴雨地区地形图的快速更新,它甚至是唯一可行的解决之道。由于有关几何处理、辐射定标等基础问题没有很好地解决,影响了这一技术在我国的大规模应用及产业化进程。本文致力于解决SAR影像的几何问题及与地形有关的辐射问题,对合成孔径雷达图像的几何特性作了系统深入的研究,以对构像方程的分析及推导为中心,研究并解决了包括地理编码、目标定位、影像模拟、利用控制点进行空间轨道精确重建、地形辐射影响的消除等一系列问题。为了加强对合成孔径雷达图像的理解,首先对合成孔径雷达成像的技术本质从数学上进行了简明阐述。从信号处理的角度,分析了脉冲压缩的工作原理,解释了匹配滤波器的构造。分析了多普勒频率的特征及其作用。从理论上推导了SAR距离向和方位向分辨率所能达到的极限值,并且指出了他们在实际中的限制。从系统的角度,分析了SAR距离向和方位向模糊度的限制。构像方程是所有几何处理的基础。为推导了SAR构像方程,在定量分析了地球摄动力对卫星轨道影响的基础上,提出了一套改进的SAR轨道参数模型,与国外已有的模型相比,该模型更加简洁而且具有极高的精度。从距离方程和多普方程出发,推导了建立在轨道参数和成像处理参数基础上的SAR构像方程。将SAR几何校正问题分解为地理编码与空间定位两方面,以构像方程为基础,通过牛顿迭代法解决了SAR地理编码问题。对空间定位的重要意义进行了分析,因为它是一个二维平面到三维空间的反演问题,实际解算非常困难,本文提出了一种巧妙的解决方案。精确地确定轨道数据,对于几何校正处理,SAR干涉测量和SAR立体测图都有着重要的意义。以构像方程为基础,提出了通过至少5个地面控制点进行轨道精确测量的算法。系统地介绍了SAR影像模拟技术,指出了模拟影像与真实影像之间的位置差异正是采用了不准确的轨道数据的结果,提出了无须地面控制直接利用模拟影像进行精确的几何校正的详细策略。将地形对后向散射的影响归结为面积效应和局部入射角效应。为消除面积效应的影响,推导了面积归一化因子,并指出了被国外研究者使用的几种不同面积归一化因子之间的关系;为消除局部入射角影响,提出了以局部入射角的线性函数表达的后向散射模型,并据此生成了校正函数。在此基础上,给出了消除地形对后向散射影响的算法步骤。最后,对本文的关键算法进行了编程实现,并以ERS和RADARSAT SAR影像为实验数据,进行了大量详细的实验,验证了本文提出的基于构像方程的几何校正和轨道精确测量、SAR影像模拟以及应用于几何校正、地形对后向散射影响的消除等算法的正确性,并对结果进行了分析。更多还原

【Abstract】 As an advanced Earth observation technology emerging in the 20th century, Synthetic Aperture Radar was broadly applied in the field of land cover mapping, ecology and agriculture, solid Earth science, hydrology, sea ice, because of its all weather day/night imaging capability and penetrating ground surface to some extent capability. In the future, Synthetic Aperture Radar will play a more important role in broader range, with the outcoming high resolution, multipolarization, multiband and optimal interferometry configuration.In China, SAR remote sensing technology has an extensively potential application market. Moreover, SAR maybe the only viable alternative for topography mapping in western China and topography map rapid updating in part of southern China characterized by persistent cloud-cover. But some issues concerning SAR geometric processing and radiometric calibration are not well resolved, which hinder this advanced technology from being applied on a large scale and being industrialized.The goal of this dissertation is to deal with the geometry-related issues and topography induced radiometric distortion in SAR images. On the basis of thorough and systematic analysis of geometric characteristics, the imaging equation of SAR images is developed, then, the issues concerning SAR geocoding, localization, image simulation, orbit precise determination by means of GCPs and terrain influences on backscatter and attempts to their correction. , are investigated in detail.In order to understanding SAR imagery profoundly, the principles of SAR forming process are formulated. How the pulse compression works and what is matching filter are analyzed in the viewpoint of signal processing. The Doppler frequency and its property are well addressed. The ultimate value of azimuth resolution and range resolution of SAR images are theoretically deduced. Then, the range ambiguity and azimuth ambiguity are considered.Imaging equation is the basis of geometric processing. After investigating the effect of Earth pertubatory forces on satellite orbits, a different parameters set describing orbit from I. Tannous et al (1994) and H. Rantakokko et al (1999) is proposed, which is the optimal trade-off between accuracy and computational simplicity. Then, the SAR imaging equation based on orbit parameters and SAR processor parameters are developed. SAR image geocoding, which is a transformation from object space to image plane, and SAR localization, which is a transformation from image plane to object space, consequently inversion of geocoding, are explored. The iterative algorithms to resolve localization and geocoding problem by means of imaging equation are presentedPrecise determination of SAR orbit is significant to SAR geocoding, as well as interferometric SAR processing and stereo SAR processing. A orbit refinement algorithm using ground control points is proposed, which linearizes the imaging equations and solve the orbit parameters under least square principle as space resection of aerophography.SAR simulation is a two-step process, one is geometric calculation of pixel position corresponding to DEM cell, and is similar with geocoding, the other is backscatter simulation. Frequently used backscatter models are reviewed. The difference between real and simulated images provides the control to refine orbit model. A hierarchical matching strategy between real and simulated SAR images is proposed and applied to automatic SAR geocoding.Terrain induced radiometric distortions have to be corrected when performing absolute radiometric calibration and relating backscatter with biomass, moisture and land cover classes. These distortions are categorized pixel scattering area effect and local incidence angle effect. To correct pixel scattering area effect, the area normalization factor is developed, to correct local incidence angle effect, a backscatter model formulated as a linear function of local incidence angle is presented, and then correction function. As an end, the correction algorithm of terrain induced radiometric distortions is developed.Last but not least, all the developed algorithms are programmed using c++. Utilizing ERS SLC data and RADARSAT SGF data, four tests are arranged, respectively about imaging equation based geocoding, orbit refinement using Ground Control Points, SAR simulation and its application to automatic geocoding, terrain induced radiometric distortion correction. The results of the tests are analyzed and concluded.更多还原