基于摄影测量技术的房屋提取方法——以中国西部地区乡村为例

发布时间：2018-03-19 14:17

  本文选题：房屋分布　切入点：可见光遥感　出处：《地震地质》2017年04期 　论文类型：期刊论文

【摘要】：房屋的分布位置、面积和高度等信息对城乡规划、地震应急和减灾等方面具有非常重要的作用。利用星载或机载可见光遥感图像进行地物信息自动提取,是高效快速获取大面积房屋数据的重要手段。但是,当出现同物异谱或异物同谱现象,以及地物遮挡的影响时,基于光谱信息的传统地物分类或提取方法房屋提取精度较低。文中选取新疆维吾尔自治区琼哈拉峻村房屋和农田光谱信息相似的地区,利用小型旋翼无人机获取了在航向和旁向具有一定重叠度的可见光遥感图像。然后,基于摄影测量原理获得了研究区的数字表面模型(Digital Surface Model,DSM)、数字高程模型(Digital Elevation Model,DEM)和数字正射影像数据,并在此基础上通过阈值分析DSM和DEM差值提取出房屋像元及其面积和高度等信息。通过与传统监督分类法对比发现,使用摄影测量方法基于高度信息的房屋提取其用户精度和制图精度分别为88.69%和97.42%,而监督分类法提取房屋的精度分别为43.23%和85.30%,说明文中提出的方法在房屋和背景信号差别较小的区域时精度更高。
[Abstract]:The location, area and height of buildings play an important role in urban and rural planning, earthquake emergency response and disaster mitigation. It is an important means to obtain large area house data efficiently and quickly. However, when the phenomenon of isomorphism or foreign body isospectral, and the influence of object occlusion, The traditional methods of classification or extraction of ground objects based on spectral information are of low precision. In this paper, the areas with similar spectral information between houses and farmland in Qiongharjun Village, Xinjiang Uygur Autonomous region, are selected. Using a small rotor-wing UAV, the visible light remote sensing images with a certain overlap in the heading and the side direction are obtained. Based on the principle of photogrammetry, digital surface model (DSM), digital elevation model (DSM) and digital orthophoto image data (DSM) and digital orthophoto image (DSM) are obtained. On this basis, the information of house pixel, its area and height are extracted by threshold analysis of DSM and DEM difference. The accuracy of house extraction based on height information using photogrammetry is 88.69% and 97.42, respectively, while that of supervised classification is 43.23% and 85.30, respectively. The area with less difference is more accurate.
【作者单位】： 中国地震局地质研究所;
【基金】：“大中城市地震灾害情景构建”重点专项(2016QJGJ14) 中国地震局地质研究所基本科研业务专项(IGCEA1522)共同资助
【分类号】：P23;P315.9
，

本文编号：1634629