主成分估计用于解算卫星线阵CCD影像外方位元素

在航天摄影测量卫星线阵CCD影像的外方位元素解算中,经常产生严重的病态性问题,如果采用最小二乘原理解算,其解明显偏离真值,甚至无法求得外方位元素,在分析以往解决办法的基础上,引入主成分估计来解算,并根据外方位元素解算问题的实际,提出了确定主成分估计中偏参数的3种方法,实验证明该方法稳定、有效。
     

数字近景影像上标志点精确定位方法研究

本文研究数字近景图像上标志点的精确定位问题,具体围绕两张桥梁悬索模型数字近景图像进行人工标志点的定位研究。在比较现有标志点定位方法特点的基础上,提出了一种新的近景标志点定位的方法。针对所研究的图像,比较各方法特点并进行单个标志点实验,得出各方法的优缺点。利用可行的两种方法对两张近景影像上的所有标志点定位,通过直接线性变换解算像片参数和检验点坐标,根据两种定位方法以及目视定位的解算结果,比较各方法的精度,得出结论———本文所提出的定位方法精度较高,采用该方法进行标志点定位,可提高像片解算的精度。     

星载焦面视场拼接TDI CCD相机成像质量及处理方法分析

本文分析卫星偏航角、相机积分时间、平台稳定性、姿轨误差对焦平面线阵采用视场拼接的TDI相机地面成像的影响,包括线阵、沿轨两个方向上的错位(点上的变化)与地面扫描轨迹方向上的影像畸变(线上的变化).针对CBERS-02B HR相机的卫星影像,研究地面解决偏航角引起图像错位的数据处理方法——基于地面影像的图像配准仿射变换纠正法.根据CBERS-02B HR线阵的地面配准试验,分析出了CBERS-02B卫星星上偏航角补偿精度与误差对图像质量的影响,及影像畸变产生的原因.     

虚拟CCD线阵星载光学传感器内视场拼接

以基于虚拟CCD线阵的多CCD影像重成像算法作为内视场拼接的技术手段,在对由地形起伏引起的多CCD影像的拼接误差进行理论分析和推导的基础上,提出无需数字高程模型(DEM)的虚拟CCD线阵多CCD影像重成像算法;并提出使用基于严密成像几何模型的空间前方交会的方法直接评价影像拼接对摄影测量生产的精度影响。研究表明,在虚拟CCD"安装"位置与真实CCD位置偏差不大的情况下,使用成像区域的平均高程进行拼接即可满足几何无缝拼接的需求;而在虚拟CCD"安装"位置与真实位置的偏离超过限差的情况下,可以用航天飞机雷达地形测绘DEM(SRTM-DEM)等一定精度的DEM数据校正地形起伏引起的拼接误差。使用先进陆地观测卫星(ALOS)卫星全色遥感立体测绘仪(PRISM)传感器三线阵影像作为实验数据,对前视、下视和后视的多CCD影像分别进行拼接。对拼接线的判读结果表明,影像拼接效果良好。另外,分别对拼接前和拼接后的前视、后视影像选取同名点进行空间前方交会,拼接后影像的空间前方交会精度与拼接前影像的空间前方交会精度一致。本文方法,可以无需DEM进行ALOS PRISM的内视场拼接,并使得拼接后影像的立体测图精度无损;拼接误差分析方法也可以在航空相机影像拼接中推广。     

航天线阵CCD传感器内方位建模与优化

线阵CCD传感器在航天摄影测量中已获得了广泛的应用,但卫星发射上天后由于多种因素,CCD、镜头的各项参数均发生了变化,原来实验室里标定的值将不再适用。此时卫星影像定位必须顾及像点的实际偏差,否则将对定位精度产生很大的影响。本文详细分析了造成这些偏差的因素,构建了合适的内方位模型,然后以HRS影像为例对模型中的参数进行了解算并作了优化。试验表明,本文所建立的内方位模型是合理的。     

基于STK_Matlab的共形相控阵对卫星可见性分析

近年来中国在轨卫星数量与日俱增，航天地面测控资源日益紧张，地面站现有的球面共形相控阵存在极大的阵面资源浪费，针对地面测控站阵面波束资源的调度分配可以有效利用现有阵面资源，为日后数量更庞大的航天测控任务需求增加资源裕度，为此需要对中国在轨卫星相对阵面的分布情况进行可见性分析；设计了一种基于STK与Matlab互联的仿真计算方法，该方法利用中国在轨卫星实时TLE数据，通过Matlab对STK控制仿真卫星相对地面站的分布及可见性，并把仿真数据存入Matlab，进一步仿真卫星对子阵的可见性，统计对各子阵的可见性结果；仿真结果得出了球面阵上各子阵在全天时刻的可见卫星数量以及资源较为紧张的子阵分布，该结果可以为球面共形相控阵波束资源调度设计提供参考。     

基于点和直线特征的高分辨率航空单线阵影像几何纠正

针对恶劣飞行条件下,高分辨率航空单线阵影像基于POS数据几何纠正后仍存在内部畸变的问题,利用影像上丰富的点线纹理特征,开展了基于特征的高分辨率航空单线阵影像几何纠正方法研究,分析了影像段落小带来的外方位元素解算时复共线性问题,采用线角分开迭代法实现了外方位元素的求解,使得该模型能够适用于高分辨率航空单线阵影像的几何纠正.实验结果表明:此方法能够有效校正因外方位元素测量引起的航空单线阵影像几何畸变.     

基于资源三号卫星数据的数字正射影像图制作研究

资源三号卫星的成功发射填补了我国立体卫星测绘的空白,可以提供优于5m地面分辨率的遥感影像.论文提出基于资源三号卫星遥感数据的数字正射影像图（DOM）制作流程,重点对影像纠正、影像配准、影像融合、影像镶嵌等关键步骤进行研究,并通过实验验证方法的可行性.实验证明该方法具有较强的实用价值和广泛的应用前景.     

LiDAR数据与CCD影像融合算法研究

依据新型遥感载荷LiDAR数据高程精度高、CCD面阵影像平面精度高的特点,本文提出一种基于点特征的激光点云与CCD影像高精度匹配的方法,研究了立体像对与激光点云联合同名点搜索算法和影像特征点与离散激光点迭代匹配求解算法,实现了一个像对区域的影像与点云高精度配准,获得了三维彩色点云影像。得到的影像既具光学影像的颜色纹理信息又具有LiDAR数据的高程信息,提高了地物目标识别解译的可靠性和量测的精度,也为更进一步的地物分类提供更为精确的原始数据。     

基于NARX神经网络辅助组合导航方法研究

为解决短时全球导航卫星系统(GNSS)失效造成车载组合导航系统导航精度下降的问题,提出一种NARX神经网络辅助的组合导航方法。对神经网络辅助导航的原理进行了分类,并分析了神经网络可利用的输入输出信息,提出一种根据惯性测量单位(IMU)测量信息和惯性导航解算信息对GNSS位置速度增量进行预测的方法。通过实测数据实验验证了方法的有效性,GNSS失效60 s期间,导航最大位置误差5.1 m、最大速度误差0.15 m/s。     

End-to-end simulation model of rover-based hyperspectral remote-sensing systems: application to VNIS

The simulation of remote-sensing hyperspectral images has various applications such as the design of future hyperspectral imaging systems, understanding of the image formation process, development and validation of data processing algorithms, and optimization of the instrument imaging mode. For incomplete understanding of the lunar surface and the wide environmental differences between earth and moon observation, imaging systems for lunar observation cannot be tested in their exact working environments before launch. In these cases, simulation of lunar hyperspectral images can be used as a powerful tool to analyse the imaging process on the moon. The VIS–NIR imaging spectrometer (VNIS) aboard the Chang’e-3 (CE-3) lunar rover is used to perform in situ mineral detection on the lunar surface, but the rover-based VNIS has some additional effects from the rover itself (e.g. the shadow caused by the rover). In this paper, a rover-based radiative transfer model has been developed, and the simulation model is able to generate realistic VNIS-like data in an automatic way under a set of user-driven instrument and illumination parameters. A realistic surface reflectance cube, as the original input for the simulation model, is provided by the interference imaging spectrometer (IIM) data of Chang’e-1 (CE-1). Several hyperspectral simulations in different illumination and observation geometries have been conducted to analyse the effects of shadow, specular irradiance, and diffuse irradiance on the imaging data. For certain illumination geometries, the simulation model can forecast image quality in different observation geometries; the model can also determine the optimal observation azimuth ranges at different solar elevation angles. Moreover, the simulation model can be used to provide test images for the rover effect elimination algorithms. These applications of the model can facilitate understanding by analysing the rover-based hyperspectral remote-sensing process and eventually obtaining high-quality images of the lunar surface.     

编队卫星相对位置测量信息自主确定月球引力场的方法

针对月球远月面引力场的数据不能直接测量而导致其测量精度不高的问题,提出了一种基于双星编队的月球引力场自主确定方法。该方法考虑到月球引力场对编队卫星相对运动的影响,通过扩展卡尔曼滤波算法对星间距离测量数据的处理,利用观测量的测量值与一步预测测量值之间的偏差来修正一步预测状态值,从而得到状态估计量,即实现了卫星自主定轨以及月球引力场的自主确定。仿真结果表明,根据对编队卫星在7 430s内的相对位置测量数据的处理,卫星轨道位置和速度确定精度分别能达到5.04m和6.07×10~(-3) m/s;月球J2项摄动常数的精度能达到9.14×10~(-8) ;月球引力场常数的精度能达到3.47×10~7 m~3/s~2。此结果能在一定程度上改善现有的月球引力场模型,为我国"嫦娥工程"提供更多的技术资料。     

不同指数法在地表水体提取中的效果比较

由于遥感技术能够快速、高效地获取地表水体的时空分布特征,目前基于影像提取内陆水体的方法很多,但针对不同类型的水域,哪一种方法提取效果更好,是值得探讨的问题。以天健湖、须水河和黄河郑州段3个水域为研究对象,基于GF-2,Landsat 8,SPOT5卫星影像,采用水体、植被指数法等几种方法提取水域部分。通过分析提取效果,得出:对于水体较浅的天健湖,无论是GF-2还是Landsat 8影像,提取效果较好的方法是水体指数法,提取效果较差的均为单波段阈值法;对于相对较深的须水河,无论是GF-2还是Landsat 8影像,提取效果较好的方法是植被指数法,提取效果较差的均为单波段阈值法;对于含沙量较大、有细小水体的黄河水域,提取效果相对较好的是水体指数法,较差的是单波段阈值法和植被指数法。表明:在基于影像提取水体时,首先应弄清水域的情况,以采用相应的遥感指数。     

高分辨率遥感卫星影像的河流提取方法研究

为了克服单纯采用光谱信息提取河流的缺陷,利用高分辨率遥感影像突出的高分辨率的特性提出一种综合影像中光谱、纹理、几何特性等多特征联合提取河流的方法。该方法分别对河流水体的光谱特征、纹理特征及河流几何形状进行描述,选取特征参数,构造综合特征矩阵,利用均值聚类分割最终得到河流目标。通过对真实高分辨率遥感影像Worldview1影像进行的实验验证了该方法的高精准性及快速性。     

Synthetic satellite imagery for current and future environmental satellites

During the next decade, data from a new generation of US geostationary and polar orbiting satellites will become available. To prepare for these data, representative imagery of these satellites is desirable. Two independent methods have been developed to create imagery from future satellites before they are placed into orbit. One method uses data from current operational and experimental satellites. Data obtained this way are referred to as simulated imagery. Another method generates satellite imagery by using numerical models. Data obtained by this method are referred to as synthetic imagery. Each method has some weaknesses that can be overcome by using both methods together. Synthetic imagery for two future US sensors is introduced in this paper. Emphasis is placed on a severe thunderstorm event.     

Geometric cloud heights from Meteosat

Atmospheric Motion Vectors (AMVs) constitute an integral part of the global observing system. The heights of these vectors are determined by the target temperature, which, in the case of semi-transparent or broken clouds, is corrected for the upwelling radiance from scenes below the target. Despite the success of these methods, the attributed heights are still the largest single source of error of the AMVs. In order to improve the existing methodologies to determine the height of the AMVs alternative methods have to be employed. The potential of geometric cloud height estimates, where two satellites view the same cloud, has already been demonstrated with geosynchronous satellite data only as well as together with data from polar orbiting satellites. In these studies the cloud targets have been selected manually and the analysis has generally been based on imagery data from the visible channels. In this paper we take the methods further by applying them to geostationary infrared imagery data, using a fully automated procedure. The reason for exploiting infrared data is the capability of continuous observations also during night-time. The results presented demonstrate the stereo method as a verification and analysis tool for satellite cloud top temperature height retrievals and that it can be applied to 5-km resolution infrared imagery data from the geostationary satellites Meteosat 5 and Meteosat 7. A further analysis combining Meteosat data with imagery data from a polar orbiting satellite shows that the automated geometric technique can be extended beyond the range of overlapping geosynchronous observations, using an asynchronous stereo technique.     

Cloud/snow recognition for multispectral satellite imagery based on a multidimensional deep residual network

Cloud/snow recognition technology for multispectral satellite imagery plays an important role in resource investigation, natural disasters, and environmental pollution. Traditional feature based classification methods cannot make full use of the effective features and multispectral optical parameters of satellite imagery; the precision of cloud/snow recognition is not good enough. Although deep convolution neural network (CNN) can extract features effectively, it faces training gradient diffusion and model degradation, which lead to a low accuracy in classification. In order to solve this problem, an improved deep residual network with multidimensional input is proposed for the cloud/snow recognition. The multidimensional deep residual network (M-ResNet) can effectively extract the image features and spectral information of satellite imagery. The multispectral satellite imagery is divided into cloud/snow-free, cloud only, snow only and cloud/snow mixed using the proposed method. The experimental results of HuanJing-1A/1B (HJ-1A/1B) satellite imagery in China show that the M-ResNet performs a good distinction for the four kinds of images. The accuracy of the classification is higher than support vector machine (SVM), random forest, convolution neural networks, and multi-grained cascaded forest (GcForest).     

高分辨率卫星影像车辆检测研究进展

高分辨率卫星遥感技术具有在更小的空间尺度上探测地表目标的能力,利用其影像数据进行车辆检测已成为新的研究热点。在概述遥感影像车辆检测研究现状的基础上,对车辆目标影像特征及车辆检测过程进行了探讨;将车辆检测方法分为利用光谱/几何结构特征的基本检测方法和综合运用多种特征的智能化检测方法,并详细叙述了多种车辆检测方法的原理与适用性以及车辆提取中的关键技术。通过分析发现：结合多特征的机器学习和面向对象的车辆检测方法更适合较复杂环境下的车辆检测。     

Georeferencing of UK DMC stereo-images without ground control points by exploiting geometric distortions

This article describes a new method for the georeferencing of UK-DMC imagery that does not require ground control points (GCPs). The proposed method utilizes satellite ancillary data, and the inter-imager offsets to determine the geolocation of individual pixels. The major step involved is the direct georeferencing of each pixel using satellite GPS and attitude sensor observations. The known separation between the sensors will allow us to determine the geolocations of all pixels that are taken at the same time using the same exterior orientation parameters. Traditional methods for georeferencing use GCPs, which are expensive and time-consuming tasks. Moreover, the traditional method is not suitable for a pushbroom imager because every scan line has a different set of exterior orientation parameters. Therefore, we propose a direct georeferencing approach without GCPs. The major source of error in direct georeferencing is the error in attitude measurements. The reason for this error is considered to be the thermo-elastic effects on the satellite, which affect the sensors’ positioning, causing deformation in the images. These effects have been modelled as a transformation matrix that describes the extent of deformation in the imagery, and is estimated by exploiting the geometric distortions in stereo Earth images. For this purpose, a mathematical model has been developed to demonstrate how inter-image offsets have been introduced into the imagery and affected by thermal deformation. The mathematical model is based on the sensor configuration of UK-DMC satellites. The model has been further inverted to extract the thermal deformation at a given row and column offset. The thermal deformation matrix has been found to mitigate the pointing error up to 1 km. The accuracy of the thermal deformation estimates is highly dependent on the accuracy of image offsets. The accuracy of image offsets is dependent on several factors, which include the image registration method, window size, along-track separation between the sensors, satellite attitude, and resolution of the sensors.