摘要
高边坡地质灾害分析与防治,需要在较大范围采集结构面的几何信息。但是,高边坡(尤其是临江崖壁),由于人员抵近接触式测量危险大、成本高,效率低,而远距离采用光学或遥感方法难以获得足够精度信息,导致结构面产状、迹线等信息难以准确、快速地获取。通过工程现场测试给出无人机采集高边坡虚拟多目数字图像的方法与步骤,以单相机虚拟多目视觉方法为基础,对临江高边坡拍摄目标进行三维点云重构,并利用K-means聚类算法和特征点收缩算法进行结构面及其迹线的识别和提取。使用该方法对四川乐山金口河峡谷边坡进行测试,结果表明此法具有较高的速度和准确性,在与边坡相关的施工中有很大的应用前景。
The analysis and prevention of geological disasters in high slopes require the gathering of geometric information of discontinuities in large areas.However,in high slopes(especially riverside high slopes),manual measurement is highly dangerous,costly and inefficient,and long-distance measurement using optical or remote sensing methods has low accuracy,making it hard to extract information such as discontinuity orientation and traces accurately and quickly.Through field tests in actual engineering,a method of gathering multi-vision virtual images of high slopes using drones is proposed.Based on virtual multi-vision algorithm of a single camera,the 3 D point cloud of the photographed riverside high slope is reconstructed.K-means clustering algorithm and feature point contraction method are used to recognize and extract discontinuities and their traces.Testing of this method on a slope in Jinkou River canyon in Leshan,Sichuan reveals that the method is high in speed and accuracy,and has big implications in slope-related engineering.
作者
易小凯
武威
张可珅
徐博
李华明
刘剑
何国华
Yi Xiaokai;Wu Wei;Zhang Keshen;Xu Bo;Li Huaming;Liu Jian;He Guohua(Tongji University,Shanghai 200092,China;State Key Laboratory of Rail Transit Engineering Informatization(FSDI),Xi’an 710043,China;China Railway First Survey and Design Institute Group Co.,Ltd.,Xi’an 710043,China;Sichuan Lehan Expressway Co.,Ltd.,Leshan 614099,China;Zhong Dian Jian Ji Jiao Expressway Investment Development Co.,Ltd.,Shijiazhuang 050011,China;Guizhou Expressway Group Co.,Ltd.,Guiyang 550003,China)
出处
《土木工程学报》
EI
CSCD
北大核心
2022年第S02期139-148,共10页
China Civil Engineering Journal
基金
国家自然科学基金(41902275,U1934212,41827807)
关键词
无人机
高边坡
多目视觉
三维重构
结构面识别
几何信息提取
drones
high slopes
multi-vision
3D reconstruction
discontinuity recognition
geometric informationextraction
