摘要
边坡灾害调查是及时发现边坡灾害隐患、预警边坡突发事故、避免重大人员伤亡和财产损失的重要手段。为克服传统人工边坡调查方法效率低、风险高、难度大等缺陷,提出了基于无人机倾斜摄影的公路边坡三维重建和灾害识别方法。引入运动恢复结构(Structure from Motion,SfM)和多视图立体匹配算法(Multi View Stereo,MVS),利用无人机多视角序列影像重构了公路边坡三维实景模型;利用基于多尺度模型与模型点云比较算法(Multiscale Model to Model Cloud Comparison,M3C2)的三维点云数据变化检测技术,实现了公路边坡滑坡、坍塌、落石等灾害场景的自动识别;通过边坡现场验证试验,证实采用无人机倾斜摄影方法构建的边坡三维实景模型的单方向精度均优于2.0 cm,并成功识别用于模拟边坡变化的最小尺寸为8.0 cm×9.0 cm×13.0 cm在内的纸箱位置、形状及尺寸;将该技术应用于某公路边坡工程地质灾害调查,成功识别出一处塌陷区域及隆起区域大约为65.0 m×15.0 m的边坡滑坡病害,构建的三维实景模型精度达到厘米级水平。研究结果表明:基于无人机倾斜摄影的公路边坡灾害调查方法适用于城市道路、公路等边坡自动化检测,尤其适用于陡峭、危险性高、难以抵达的边坡智能化检测,具有较强科学研究和工程实用价值。
Slope disaster investigation is an important work to timely discover hidden dangers of slopes,early warn slope accidents,and avoid major casualties and property damage.In order to overcome the shortcomings of the traditional manual survey such as low efficiency,high risk and difficulty,this paper presents a novel method for highway slope three-dimensional(3 D)reconstruction and disaster Identification using unmanned aerial vehicle(UAV)-based oblique photography technique.A 3 D reconstruction technique was introduced using Structure from motion(SfM)and multi-view stereo(MVS)algorithms.3 D realistic models of highway slopes could be reconstructed using UAV-acquired multi-view images.An algorithm based on Multiscale Model to Model Cloud Comparison(M3 C2)was used for change detections of 3 D point cloud data.Disaster scenes of highway slope such as landslide,collapse and rockfall could be identified automatically.A verification experiment was conducted to assess the accuracy of 3 D realistic model in each direction.All of accuracies in three directions of the slope model are less than 2.0 cm.The positions,shapes and sizes of cartons,including a set of small cartons with sizes of 8.0 cm×9.0 cm×13.0 cm,were successfully identified,which were used to simulate slope changes.The proposed method was applied to a geological disaster survey of a highway slope project.A landslide disaster with a collapsed and raised area of approximately 65.0 m×15.0 m was successfully identification,and a 3 D model of the highway slope with a centimeter level accuracy was established.In conclusion,the slope survey method based on the UAV oblique photography technique is suitable for slope automation detections of both highways and urban roads,especially for steep,dangerous and hard-to-reach slopes.The proposed method has strong scientific research and engineering practical value.
作者
余加勇
薛现凯
陈昌富
陈仁朋
何旷宇
李锋
YU Jia-yong;XUE Xian-kai;CHEN Chang-fu;CHEN Ren-peng;HE Kuang-yu;LI Feng(Key Laboratory of Building Safety and Energy Efficiency of Ministry of Education,Hunan University,Changsha 410082,Hunan,China;College of Civil Engineering,Hunan University,Changsha 410082,Hunan,China)
出处
《中国公路学报》
EI
CAS
CSCD
北大核心
2022年第4期77-86,共10页
China Journal of Highway and Transport
基金
国家重点研发计划项目(2016YFC0800207)
长沙市科技计划项目(kq1907110)
湖南省水利科技项目(XSKJ2021000-46).
关键词
道路工程
灾害识别
三维模型重建
无人机倾斜摄影
road engineering
disaster identification
3Dmodel reconstruction
UAV oblique photography
作者简介
余加勇(1976-),男,湖北孝感人,副教授,工学博士,E-mail:surveying@hnu.edu.cn。
