利用无人机影像进行滑坡地形三维重建

3D Reconstruction of Landslide Terrain from UAV Images

针对滑坡表达研究的需要,提出了一种基于无人机影像序列的滑坡地形全自动鲁棒三维重建方法。以某滑坡区为试验场地,利用无人机搭载小型数码相机获取滑坡影像,辅助飞控数据建立的影像拓扑关系,依据计算机视觉原理,对滑坡地形进行了全自动三维建模,生成了包含颜色信息的三维点云数据,构建了滑坡体数字表面模型(DSM)。结果表明:该方法可快速实现滑坡三维地形精细测绘,有效降低作业成本和劳动强度;所建模型可准确表达滑坡体的空间分布特征,为正确分析评价滑坡稳定性提供了有力支持,尤其适合在困难山区或潜在危险地区的滑坡遥感动态监测。

To meet the demand of expression research for landslide,a fully automated and robust reconstruction approach form UAV images which can generate 3- D landslide terrain was put forward. This method was applied to a landslide affected area in Shandong successfully. Depending on the established image topology relationship with the flight- control data acquired by the UAVs,the proposed computer vision method can easily enables digital surface model of landslide to be built based on generated 3D color point clouds from images obtained by UAV with a small digital camera. The experiment showed that the presented method can be used for landslide terrain fine mapping,effectively reduce the operation cost and labor intensity. Moreover,the spatial distribution characteristics of landslide body can be accurately expressed through reconstructed model,which can provide powerful support for the correct analysis and evaluation of landslide stability,especially suitable for remote sensing of landslide dynamic monitoring in the difficult alpine terrain or potential dangerous areas.