作　　者： (王书涵）; (张晓丽）; (杨铭）; (刘会玲）;

机构地区： 北京林业大学林学院、精准林业北京市重点实验室、省部共建森林培育与保护重点实验室,北京100083

出　　处： 《中南林业科技大学学报》 2017年第10期22-29,共8页

摘　　要： 山区高低起伏的地形容易影响样地三维坐标的采集精度,从而限制了高分遥感在山区林业上的广泛应用。以鹫峰林场为研究区,提出了采用全站仪和手持GPS结合遥感图像协同对样地进行定位的方法,并提出了坐标修正方案:即首先运用手持GPS结合全站仪采集样地及单木的绝对大地坐标;然后利用大比例尺地形图采用基于控制点、RPC的两种正射校正方法对覆盖研究区的QuickBird影像进行正射校正,RMSE多光谱控制在1个像元之内,全色图像控制在10个像元之内;并以样地为例分析了样地及单木的匹配精度。研究结果表明:本方案能快速准确地建立样地中单木与高分影像的联系,其中基于控制点的图像正射校正法的匹配精度最高,相对提高幅度为92.6%,而基于RPC的提高幅度70.05%,即E_(ControlPoint)>ERPC。研究结果表明,对应用于地形复杂的山区的高空间分辨率的遥感影像正射校正是不可缺少的步骤,而本文提出的方案能实际应用于山区的高分遥感样地定位,具有推广利用的价值,并为山区的高分遥感进一步的应用打下基础。 Extensively applying of high resolution image was restricted because dramatical affects were caused by the rugged topography of mountain area to the geometric rectification of high resolution remote sensing imagery and coordinate collection with high precision. The forestry research region in remote sensing is often rugged topography of mountains. How to link precisely and accurately the actual land features and image information has long been the concerned and discussed issues by researchers. Jiufeng national forestry park was selected as the study area. One Chinese pine （Pinus tabulaeformis） field sample plot and Chinese arborvitae （Platycladus oriental） field sample plot B2 and B3 were selected respectively. The synergy method combining with remote sensing and total station and Handheld GPS was proposed as well as the revised scenario of coordinate. First the true value of ground objects was collected using total-station and handheld GPS, including the trees in plots and angular points. The original QuickBird multispectral and panchromatic image were respectively orthorectified using two types of orthorectification methods： method based on control points identified from topographic map and RPC method. The RMSE of multispectral image and panchromatic image is within 1 pixel and 10 pixels respectively. Road intersections, houses and other obvious features in the 1：2000 topographic map were selected as the features where the GCPs are, The topographic map surveyed by Beijing institute of surveying and mapping in Oct 2004 has been geometrically corrected. Both methods of process above were conducted in ENVI 5.0. First, feature points were selected near the sample plots whose coordinates were obtained using differential GPS system. Matching accuracy between 3D coordinates of feature points and feature positions in the QuickBird image was analyzed. The results shows that matching accuracy of orthorectification based on control points reached the highest level with 92.6% increased range and RPC orthorectif

关 键 词： 快鸟影像 高空间分辨率遥感图像 全站仪 手持 正射校正