基于GeoEye-1卫星遥感影像几何模型解算及三维重建精度分析

发布时间：2018-10-10 16:51

【摘要】：21世纪以来,受益于计算机科学和材料科学的飞速发展,高分辨率遥感卫星技术得到了极大的提高；与此同时,测绘大比例尺地形图依赖于航空摄影测量的限制被打破,航天摄影测量开始广泛的应用于各大中比例尺地形图的测制。而且由于高分辨率遥感卫星卓越的数据采集和传输能力,且不受航空管制的影响,高分辨率遥感卫星技技术在国民经济建设等诸多领域发挥着越来越重要的作用。高分辨率遥感卫星往往采用与太阳同步轨道,较高的飞行高度和较高的分辨率决定了其长焦距、窄视角的物理特性；而且,高分辨率遥感卫星大都使用线阵推扫式CCD传感器,成像原理及其复杂,且传感器方位元素之间往往存在较强相关性；再者,各卫星商业公司出于技术保密的考虑,一般不向用户提供影像的严密传感器模型参数,基于此,有理函数模型(Rational Function Model, RFM)被广泛应用于高分辨率遥感卫星影像定位中,以替代严密传感器模型模拟像方坐标到地面物方坐标的映射关系。针对高分辨率卫星影像传感模型定位特征,探讨了有理函数模型系数的解算方法,研究了分别基于地形相关和地形无关方案解算有理函数模型的模型推导及计算方法,并且根据其误差特性,重点研究了两种补偿优化模型,并通过实验验证其理论和精度。地形无关方案以严密传感器模型通过构建像空间格网来拟合计算有理函数模型系数,具有较高的定位精度,但是,由于缺少地面控制信息的改正,往往出现系统性偏差。本文在像空间间接优化模型的基础上通过构建三维空间格网实现了像空间间接优化模型到直接模型的转换,并且通过实验验证了该模型在定位及三维重建方面具有较高的精度。地形相关方案是以地面像元控制点作为解算单元,当地面相对平坦时,可得到较高的有理函数模型系数解算精度。影像区域高差起伏较大时,受地面控制点数量和分布情况的影响,在远离解算控制点时,误差往往较大,且呈现随机性,很难进行改正。本文从有理函数数学特性着手,通过对模型“简化—再复原”方法着重研究了有理函数模型的系统规律,通过对模拟数据的研究,提出了基于有理函数模型的系统补偿办法,经实验检验,得到了较高的精度。此外,通过基于有理函数模型交会过程的研究,推导了基于有理函数模型的三维重建技术,使用GeoEye-1卫星遥感影像像对分别在RPC解算及优化的基础上完成了重建精度的评定,并在此基础上完成了立体测图；最后,在外部高精度DEM的基础上实现了对测绘地形图的精度评定。
[Abstract]:Since the 21st century, with the rapid development of computer science and material science, the technology of high-resolution remote sensing satellite has been greatly improved; at the same time, the limitation that large-scale topographic maps depend on aerial photogrammetry has been broken. Spaceflight photogrammetry has been widely used in topographic mapping of large and medium scale. Because of the excellent data acquisition and transmission capability of high resolution remote sensing satellite and not affected by aviation control, the technology of high resolution remote sensing satellite plays an increasingly important role in many fields such as national economic construction. High-resolution remote sensing satellites often use sun-synchronous orbit, which determines the physical characteristics of their long focal length and narrow angle of view due to their high flying altitude and high resolution. Moreover, most high-resolution remote sensing satellites use linear push-sweep CCD sensors. The imaging principle is complex, and the sensor azimuth elements often have strong correlation. Furthermore, the satellite commercial companies generally do not provide users with strict sensor model parameters for the sake of technical confidentiality. The rational function model (Rational Function Model, RFM) is widely used in high resolution remote sensing satellite image positioning to simulate the mapping relationship between the image square coordinates and the ground object coordinates in place of the rigorous sensor model. According to the localization characteristics of high resolution satellite image sensing model, the calculation method of rational function model coefficient is discussed, and the model derivation and calculation method of rational function model based on terrain correlation and terrain independence scheme are studied respectively. According to its error characteristics, two kinds of compensation optimization models are studied, and their theory and accuracy are verified by experiments. The terrain independent scheme uses the strict sensor model to fit and calculate the rational function model coefficients by constructing the image space grid, which has high positioning accuracy. However, due to the lack of correction of the ground control information, there is often a systematic deviation. On the basis of the indirect optimization model of image space, this paper realizes the transformation from the image space indirect optimization model to the direct model by constructing the three-dimensional space grid, and the experiment proves that the model has a high accuracy in positioning and 3D reconstruction. The topographic correlation scheme takes the control point of the ground pixel as the solution unit. When the ground is relatively flat, the accuracy of the coefficient solution of the rational function model can be obtained. When the height difference of the image region fluctuates greatly, it is affected by the number and distribution of the ground control points. When it is far away from the solution control point, the error is often large, and it is random, so it is difficult to correct. Based on the mathematical characteristics of rational function, this paper focuses on the systematic law of rational function model through the method of simplification and reversion, and puts forward a system compensation method based on rational function model through the study of simulation data. The experimental results show that the accuracy is high. In addition, through the research of the intersection process based on rational function model, the 3D reconstruction technology based on rational function model is deduced, and the reconstruction accuracy is evaluated by using GeoEye-1 satellite remote sensing image pair on the basis of RPC solution and optimization, respectively. Finally, the precision evaluation of topographic map is realized on the basis of external high precision DEM.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP751

【参考文献】