中文摘要：

低空无人机以其机动、快速、经济等优势,在灾害应急事件中逐渐发挥作用,而灾场三维重建也因其突破了常规无人机遥感无法快速提供三维空间信息的局限,在灾害测量中的地位日益凸显。该文面向灾害测量需求,探索一种基于低空影像和无人机自身飞控数据的灾场三维重建方法。在系统规划与平台设计的基础上,选取某滑坡区进行试验。依据计算机视觉原理,通过特征提取、影像匹配、运动与结构重建等实现了相机位置及姿态的恢复,并通过地理注册最终完成了灾场影像三维重建。通过布设地面标识点并与其GPS RTK测量坐标比对,证实灾场重建模型的相对误差低于4‰。

英文摘要：

The increasing demand for Unmanned Aerial Vehicles（UAVs） is typically attributed to the low manufacturing and operational costs,flexibility of the platforms in disaster response missions.As the limitations of conventional UAV remote sensing in spatial information collections,disaster scene reconstruction plays an important role in disaster surveying.To meet the demand of disaster surveying,this paper developed a relatively accurate and low-cost 3D-reconstruction method to generate 3D disaster scene from UAV images supported with flight-control data.Based on a UAV system planning and platform design for disaster surveying,a landslide affected area in Shandong is selected for test.For disaster scene reconstruction,a method was proposed based on computer vision to recover camera′s positions and postures through feature extraction,image matching and Structure from Motion（SfM） reconstruction.The UAV′s flight-control data was applied to conduct the geo-registration for the whole point cloud model.To verify the accuracy of the reconstructed 3D scene,GPS-RTK（Real Time Kinematic） measured ground control points were used for absolute error and relative error analysis.The results indicated that a relative error less than 4‰ was achieved for the reconstructed 3D scene.It demonstrates that the proposed UAV image reconstruction based approach for disaster surveying is feasible and effective.