Generation of geometrically and radiometrically terrain corrected SAR image products

Terrain undulations affect the geometric and radiometric quality of synthetic aperture radar images. The correction of these effects becomes indispensable when quantitative image analysis is performed with respect to the derivation of geo- and biophysical parameters. The paper presents a rigorous approach for geometric and radiometric correction of SAR images. Using a digital elevation model, the imaging geometry is reconstructed and is used to perform geometric and radiometric correction of terrain induced distortions. The importance of a stringent radiometric correction based on the integration of the image brightness is emphasized. The approach guarantees that the energy contained in the image data is preserved throughout the geocoding process. The resulting backscattering images are fully terrain corrected and can be used for further quantitative investigations and may also improve qualitative studies as e.g. land cover classifications. The technique is applicable for different sensor types and image products, including already geocoded SAR images. The effect of different resolutions of digital elevation models used for the correction of the backscattering coefficient is investigated.     

Parametric geometric correction of airborne thematic mapper imagery

The geometric correction of airborne scanner imagery has conventionally been performed in a non-parametric manner using ground control. Parametric geometric correction methods are less reliant on ground control and have potential for automation.Recently the possibility of correcting parametrically airborne scanner imagery has become viable because of advances in the availability and accuracy of devices that may be used to measure the exterior orientation of an airborne platform. This paper describes a study initiated by the U.K. Natural Environment Research Council (NERC) to develop and investigate a provisional parametric correction system. Flight trials were flown at an altitude of 1000 m and exterior orientation parameters measured by an inertial navigation system were used to produce parametrically corrected images. The results of this study demonstrate the feasibility of the parametric correction approach.     

Évaluation de deux méthodes de correction géométrique d'images Landsat TM et ERS-1 RAS dans une étude de linéaments géologiques

Exploiting the complementary characteristics of multi-source data for geological applications is particularly useful in lineament studies. Multi-source data integration requires rigorous correction methods to preserve the accuracy of the extracted information and the integrity of the composite image. This paper compares two geometric methods on Landsat TM and ERS-1 radar images: one is a photogrammetric method developed at the Canada Centre for Remote Sensing and the other is a polynomial method that is in general use. The superiority of the photogrammetric method, which accurately models the viewing geometry is demonstrated both quantitatively and qualitatively. This method achieved an accuracy of 30 m [vdot]ersus 45 m for the polynomial method on the TM images and 65 m versus 480 m on the radar images. Furthermore, it is shown that the residual geometric errors on the images rectified by the polynomial method modify the length, orientation and the number of extracted geological lineaments.     

机载多光谱扫描图象几何畸变的全自动校正

分析了引起机载多光谱扫描图象几何畸变的各种主要因素，提出并建立了利用由惯性导航系统同步记录的飞机姿态、运动和位置参数校正机载多光谱扫描图象几何畸变的方法模型。试验结果表明，与传统采用控制点的方法相比，该方法不仅具有较高的校正精度，而且可由计算机全自动完成。     

Multi-Observed Block Adjustment for Satellite ImagesWithout Ground Control Points

In order to ensure the accuracy of target area’s ground information,it is necessary to do geometric correction for remote sensing images.Geometric correction has a great influence on the application of remote sensing images.Traditional geometric correction needs ground control points.However,it is difficult to obtain ground control points in some places,such as abroad,western China and desert.To improve positioning accuracy without ground control points,multi\|overlapping block adjustment model is built.Different from traditional method,error equations are built depending on multi\|projection in image space of a single object.In this way,error equations can converge to more accurate solutions.By adjusting the rational polynomial coefficients of each image,the positioning errors in different directions are compensated to a certain extent.Thus the positioning accuracy of remote sensing images is improved.First,block adjustment model and error compensation model are built with RPC coefficients.Then,conjugate gradient algorithm is used to solve the error equations iteratively.Finally the RPC coefficients are adjusted to improve the accuracy of positioning without ground control points.The ZY\|3 data test shows that multi\|overlapping block adjustment model increase the plane positioning accuracy of remote sensing images from 19.8m to 12.9m and the method can effectively improve the absolute positioning accuracy of remote sensing images.     

星载SAR影像的定位与几何纠正

高精度几何纠正算法对SAR影像的广泛应用具有重要意义,而SAR影像纠正中的一个重要问题是建立影像定位模型。在阐述了距离多普勒定位模型的基础上,利用一景高分辨率的TerraSAR-X影像进行了地理定位实验,实验结果表明采用2次参数方程和3次参数方程描述卫星轨道位置所得的定位精度相差不大,且都能满足精度要求,为了计算方便和减少控制点的数目,通常可以采用二次参数方程来描述卫星位置。定位参数解算完成后,又采用了间接纠正法对影像进行了几何纠正,得出纠正精度不超过3个像素,表明该算法合理可行。     

全极化合成孔径雷达影像地形纠正及其在雪冰制图中的应用

针对合成孔径雷达（SAR）影像由于地形起伏引起的图像畸变问题,文章提出了基于相干矩阵的全极化SAR影像地形纠正算法,并运用于雪冰制图。该方法首先采用距离多普勒模型建立SAR成像几何模型;然后利用全极化Cloude特征分解方法对全极化SAR图像进行融合,将融合后的SAR图像与模拟图像进行配准提高SAR影像几何定位精度;最后利用投影面积归一化和极化方位角移动补偿技术对地形引起的辐射畸变进行纠正。采用中国长江源区南部唐古拉山中段冬克玛底冰川区域的C波段Radarsat-2全极化SAR数据进行验证,配准模拟SAR和原始SAR影像的控制点方位向和距离向的均方根误差（RMSE）分别为7.765和14.586个像素;经过地形纠正后的地物分类精度达80%以上。结果表明：（1）该方法能够有效消除SAR影像中几何和辐射畸变的影响;（2）地形纠正后的SAR数据在雪冰制图中具有可行性。     

基于高亮特征匹配的双视向SAR图像建筑物高度提取

由于城区场景的复杂性和SAR成像几何畸变的影响,基于单幅SAR图像的建筑物高度提取常常存在很大困难。针对这一问题,利用建筑物目标SAR成像形成的叠掩、二次散射、较强单次散射等散射机制对应的高亮特征非常典型,并且对方向性敏感的特点,提出了一种基于双视向SAR图像高亮特征与几何模型匹配的建筑物高度提取方法。首先分析了建筑物目标的SAR图像散射特征及对雷达视向的敏感性,然后构造了建筑物目标在双视向SAR图像上高亮特征几何模型,然后基于灰度均值、灰度概率分布、边界信息定义匹配函数,并利用多种群遗传算法进行优化求解,最终得到建筑物目标的高度信息。基于模拟和机载SAR图像的试验表明该方法的建筑物高度平均反演误差小于1m,可以有效提高建筑物高度反演的精度。     

Atmospheric correction of PROBA/CHRIS data in an urban environment

The Compact High Resolution Imaging Spectrometer (CHRIS) is an imaging spectrometer onboard the European Space Agency (ESA) Project for On-Board Autonomy (PROBA) satellite. However, it has been shown that CHRIS presents some miscalibration trends over the spectral region covered. This paper reports a practical procedure for the atmospheric correction of CHRIS images based on field recalibration in an urban environment. In the first stage, the spectra of surface targets are measured and used to simulate the spectral radiance at the top of the atmosphere (TOA) for each channel and to determine the recalibration coefficients of the CHRIS images. In the second stage, two methods for atmospheric correction are examined: the radiative transfer model (RTM) and the improved dark-object subtraction (IDOS) method. For comparison purposes, the empirical line method (ELM) is also evaluated. The accuracy assessment shows that the RTM with the Moderate Resolution Transmittance (MODTRAN) code provides the most accurate atmospheric correction for the multiangular CHRIS images when using the proposed procedure.     

GPU支持的SAR影像几何校正大规模并行处理

目的几何校正(又称地理编码)是合成孔径雷达(SAR)影像处理流程中重要的一个步骤,具有一定的计算复杂度,需要用到几何定位模型。本文针对星载SAR影像,采用有理多项式系数(RPC)定位模型,提出了图形处理器(GPU)支持的几何校正大规模并行处理方法。方法该方法充分利用GPU计算资源强大及几何校正过程中每个像素处理步骤一致的特点,每次导入大量像素至GPU,为每个像素分配一个线程,每个线程执行有理函数计算、投影变换、插值采样等计算复杂度高的步骤,通过优化配置dim Grid和dim Block参数,提升GPU的并行性能。该方法通过分块处理实现SAR影像大幅面处理,且可适用于多个不同分块大小。结果实验结果显示其计算加速比为38 44,为全面客观地分析GPU并行处理的特点,还计算了整体加速比,通过多个实验分析影响整体加速性能的因素,提出大块读写提高I/O性能的优化方法。结论该方法形式简洁,通用性好,可适用于几乎所有的星载SAR影像、不同的影像幅面;且加速性能明显。     

星载SAR干涉技术获取DEM及其精度分析

星载合成孔径雷达干涉（InSAR）技术是一种数据覆盖范围广、廉价、高效、方便的数字高程模型（DEM）获取方法,但在地面植被覆盖广、大气水汽含量高的地区其影像相干性随时间基线的增加迅速降低;同时,SAR卫星的轨道误差也影响DEM精度。利用ERS-1/2卫星串行模式SAR数据获取镇江地区DEM,分析了轨道误差对DEM精度的影响;根据干涉相位的统计特性,从理论上给出干涉相位噪声与相干系数和视数之间的关系。实验结果表明就干涉像对的卫星轨道误差和相位噪声而言,在小区域内DEM精度优于3.5 m。     

Planar block adjustment and orthorectification of Chinese spaceborne SAR YG-5 imagery based on RPC

Yaogan-5(YG-5), launched in December 2008, is a Chinese high-resolution spaceborne synthetic aperture radar (SAR) satellite, with a ground resolution of 3 m. However, the direct geometric positioning accuracy of YG-5 slant range images is low and so is the mosaic accuracy of the orthoimages. To improve the geometric accuracy of YG-5 orthoimages, this article proposes a strategy to calculate the rational polynomial coefficients for each SAR image and then uses a planar block adjustment method to solve for the orientation research parameters of the SAR images to achieve the orthorectification while a auxiliary digital elevation model is necessary for height constraint. Compared with the traditional orthorectification method using a single image, this strategy can ensure both uniformity in positioning accuracy of orthorectified images and high mosaic accuracy of adjacent orthoimages based on a small number of ground control points (GCPs). Tests using Chinese YG-5 satellite data over Xi’an and Xianning, China show that, using four GCPs positioned in the four corners of the test area, we can achieve independent check point plane accuracies better than ±4 m after the planar block adjustment. Finally, this article demonstrated that seamless mosaic geometry levels can be attained after the block orthorectification.     

基于目标检测的SAR图像变化检测方法

SAR图像变化检测有着广阔的应用前景,但目前的方法普遍以精确的配准为前提,使其适用范围受到限制。针对人造目标在SAR图像上的特点,提出了一种基于目标检测的SAR图像变化检测方法。通过图像中的人造目标之间相对位置关系的相似程度确定图像的变化情况,以此来降低对图像配准精度的要求。实验表明该方法在很宽的配准精度范围内都可获得较满意的结果。     

三颗高分辨率星载SAR的定位模型构建及其定位精度评价

随着TerraSAR-X,Cosmo-SkyMed和Radarsat-2这三颗高分辨率SAR卫星的成功发射,国内越来越多的用户开始通过商业渠道或通过参与SAR数据的应用示范项目免费获取到这三颗卫星的SAR数据。要很好地应用SAR数据必须首先解决其地理编码或几何校正问题,而该问题的核心是解决SAR定位模型的建立和解算方法,在此基础上就可以实现SAR影像的地球椭球校正地理编码(Geocoding of Ellipsoid Correction,GEC)处理,增强地球椭球地理编码(Enhanced Elliposid Correction,EEC)和地形校正地理编码(Geocoding of Terrain Correc-tion,GTC)或正射校正。本文研究并实现了这三颗高分辨率SAR数据的定位模型构建方法,并对GEC效果进行了评价。结果表明本文发展的定位模型构建方法是正确的,为实现这三颗高分辨率卫星SAR数据的EEC和GTC处理奠定了基础。     

风云二号气象卫星图像自动几何精校正

介绍了一种风云二号（FY-2）卫星图像自动几何精校正系统和方法。首先获得标称图坐标系下的二值化地标模板,然后通过Bayes后验概率,获得二值化FY-2待匹配图像;通过最大相关系数方法将地标模板与FY-2标称图进行匹配,获得地标偏移量;再通过质量控制获得匹配成功的地标对,实现自动的地标匹配。最后,应用DLT变换对FY-2图像进行几何精校正。大量的测试数据表明,该方法能够有效改进FY-2可见光通道图像定位精度。几何精校正后,可见光通道图像在南北方向上的定位精度提高了31.06%,东西方向上的定位精度提高了45.21%。分析指出FY-2可见光通道图像几何精校正前后,定位精度在星下点地方时正午前后均达到最佳状态,且南北方向上的定位精度优于东西方向上的精度。同时,统计分析结果表明,提出的方法成功率达到95.29%。     

基于外部DEM的InSAR图像配准方法研究

雷达图像的配准是进行雷达干涉测量(SAR Interferometry, InSAR)处理的关键,为了保证干涉相位图或形变相位图反映真实地面特性,需要雷达图像之间亚像元级精度的配准。首先综述了已有的基于外部DEM的InSAR图像配准方法的思路及其不足之处,并提出了一种全新的思路:以图像之间的相干性作为目标函数,利用搜索的方法实现了雷达成像方位向和距离向的最优时间常数的估计,从而实现雷达图像之间亚像元级配准;还进一步推导了数字高程模型(Digital Elevation Model, DEM)的误差对算法精度影响的一个更加严密的表示。结论表明,在利用精确轨道数据的情况下,美国航天飞机测地计划SRTM获得的地形数据的精度可以满足精确雷达图像配准的要求。结果表明,利用基于外部DEM算法配准雷达图像在山区和大的时间基线情况下要优于常规相干多项式配准方法,理论上可以达到百分之一个像素的配准精度。     

A simplified method of accurate geometric correction for NOAA AVHRR 1B data

There are different methods for geometric correction of NOAA AVHRR data, but these methods either pay less attention to the accuracy, or are technically complex, almost unsuitable for most users. To combine AVHRR data with other high spatial resolution satellite data, or with ancillary data in GIS, it is necessary to develop an accurate geometric correction method, which should be easy to use even for non-professional users. After analysing the pixel shape and size of AVHRR 1B data along scan line and evaluating the quality of geographical data of NOAA AVHRR 1B data set, we found that the geographical data was adequately accurate for identifying the pixel size and shape and the method was developed accordingly. The proposed method has two steps. The first step is to correct pixel distortion. The separate program performs the distortion correction, applying the geographical data of AVHRR 1B data set to assigning the value of each pixel of the desired output geographical area with given pixel size, and making logical judgment for unassigned pixel. The second step is to perform conventional polynomial transformation on the results of the first step. An application of this method is presented in the paper. To examine the precision, SAVI images derived from the geometrically corrected NOAA AVHRR band image were used to perform overlay with each other and also with a 1 :50000 river system map. A half-pixel accuracy was achieved.     

星载高分辨率合成孔径雷达影像纠正技术

为检测TERRASAR、COSMO SkyMed、RADARSAT-2等星载高分辨率合成孔径雷达影像(SAR)在土地利用调查监测中的适用性,该文针对高分辨率SAR数据和产品特性,提出了控制点选取方法,分析了不同纠正模型的应用效果。试验表明高分辨率SAR几何纠正一般需要10～15个控制点,1m聚束模式纠正中误差约3m～5m,3m条带模式纠正中误差约5m～8m,分别满足1∶1万和1∶2.5万土地调查监测几何精度要求。研究结果为构建基于高分辨率SAR数据土地利用调查监测应用技术流程和促进高分辨率极化SAR数据业务化应用奠定了基础。     

Generalized rough fuzzy c-means algorithm for brain MR image segmentation

Fuzzy sets and rough sets have been widely used in many clustering algorithms for medical image segmentation, and have recently been combined together to better deal with the uncertainty implied in observed image data. Despite of their wide spread applications, traditional hybrid approaches are sensitive to the empirical weighting parameters and random initialization, and hence may produce less accurate results. In this paper, a novel hybrid clustering approach, namely the generalized rough fuzzy c-means (GRFCM) algorithm is proposed for brain MR image segmentation. In this algorithm, each cluster is characterized by three automatically determined rough-fuzzy regions, and accordingly the membership of each pixel is estimated with respect to the region it locates. The importance of each region is balanced by a weighting parameter, and the bias field in MR images is modeled by a linear combination of orthogonal polynomials. The weighting parameter estimation and bias field correction have been incorporated into the iterative clustering process. Our algorithm has been compared to the existing rough c-means and hybrid clustering algorithms in both synthetic and clinical brain MR images. Experimental results demonstrate that the proposed algorithm is more robust to the initialization, noise, and bias field, and can produce more accurate and reliable segmentations.