-
文中主要是针对复杂地形地貌环境下的高陡滑坡群形变监测任务,开展轻小型无人机多视影像/LiDAR点云数据采集和DEM重建方案设计优化,以及不同时期重点滑坡体形变监测和分析,技术路线如图2所示。
-
文中外业数据采集时间为2021年11月28-29日,采用飞马D2000旋翼无人机平台,分别搭载了轻型的D-LiDAR2000激光载荷系统和D-OP3000载荷五镜头倾斜系统。该系统是由飞行器、载荷、地面控制站、GNSS差分定位系统及相应的数据预处理软件组成,设备主要参数如表1、表2所示。
表 1 D-LiDAR2000激光雷达模块参数
Table 1. D-LiDAR2000 LiDAR equipment parameters
Ranging
modeWave-
length/
nmRanging
precision/
cmPoint
frequency/
kHzEcho
strength/
bitsHorizontal
accuracy/
mTOF 905 ±2 240 8 0.02 表 2 D-OP3000倾斜相机参数
Table 2. D-OP3000 oblique cameras parameters
Camera
selectionSize of
sensor/mmEffective
pixelsLens focal
length/mmSONYa6000 23.5×15.6 2.430 million×5 25(down)
35(oblique)注:×5指五个相机获取的实际像素 针对研究区复杂的地形地貌环境,文中设计了安全、高效的数据采集方案,具体如下:
(1)设备准备:利用同一无人机平台下D-LiDAR2000激光雷达和D-OP3000倾斜相机数据采集设备。考虑高陡边坡邻接水库,海拔高且风力大,选取了高原无人机螺旋翼来代替普通的螺旋翼。
(2)起降点设置:研究区内的多处高陡边坡紧邻大面积水域,因此,选取至少5~6 m2、地形平坦且通讯状况良好区域作为无人机安全起降的场地。当起降点距离测区较远时,应设置不少于30%剩余电量以保证无人机安全返回。
(3)航线优化设计:因研究区地形起伏大、沟壑纵横,导致山脊线一侧GNSS信号较差。为防止长时间GNSS信号消失,设计了沿垂直于山脊线方向飞行的航线,并设置GNSS信号消失时间不超过5分钟,以保证无人机安全飞行和航次控制。
(4)飞行参数设计:为保证影像分辨率或点云密度以及DEM精度,采用变高飞行方式。
LiDAR数据采集航线设计如图3所示,主要参数设置如下:航线间距为159 m,旁向重叠度25%,平均点云密度93点/m2,飞行速度14 m/s。
基于上述方案中步骤(1)~(4),利用同一无人机平台同时采集了五镜头倾斜影像,其中,航向/旁向重叠度为80%/65%,地面分辨率为3 cm,其航线设计见图4。
-
对获取的无人机载LiDAR点云数据进行航迹和坐标解算,生成标准点云。但会存在因传感器、空中、地面干扰生成的噪声点[8-9]。因此,选取半径滤波算法对噪声点云剔除。该方法指某点所在半径r的圆域内的邻近点数小于某个阈值N,则该点视为噪声点,被剔除,否则,保留该点。如图5所示,在给定r的情况下,当N取为1时,则只有a点是噪声点,被删除;当N取2时,则a和b点都视为噪声点,被删除,而c点将被保留。
-
针对地形起伏大、高陡边坡众多的地形地貌下,D-LiDAR2000设备采用单一视角激光扫描方式进行数据采集,导致的部分高陡边坡点云数据缺失[10],使该区域的DEM失真,无法进行边坡形变测量。因此,提出将对应区域的多视倾斜影像匹配后点云与LiDAR点云融合,以改善其DEM重建的质量和完整性,实现不同时期滑坡形变分析。具体方法如下:
(1)从多视倾斜影像生成的密集点云中裁剪出包含LiDAR点云缺失区域的点集;
(2)将预处理后的两源点云数据集进行粗配准,即,实现两源数据空间基准的统一;
(3)选取迭代最邻点 (Iterative Closest Point,ICP) 算法[11]进行精配准,算法描述如下:求解待配准点云数据与参考点云数据之间的旋转参数R和平移参数T,使得两点集数据之间满足如公式(1)所示的目标函数F(R,T)值最小准则下的最优匹配[12]。
$$ {F}({R}, {T})=\frac{1}{{n}} \sum\nolimits_{{i}=1}^{{n}}\left\|{Q}_{{i}}-\left({RP}_{{i}}+{T}\right)\right\|^{2}=\min $$ (1) 式中:Pi是待配准点云;Qi是参考点云数据中对应Pi的最近点;n是最邻近点对个数;R是3×3旋转矩阵;T是3×1平移矢量。
采用上述算法将LiDAR点云与对应区域的倾斜影像生成点云精配准,通过两类点云融合实现缺失点云补偿,融合前、后结果如图6(a)、(b)所示。
从图6可以看出,经过两源点云的融合实现了对初始LiDAR点云缺失数据的补偿,改善了该区域边坡DEM重建的的完整性。为了验证融合后的点云数据质量,利用实测选取的15个GNSS控制点进行了精度验证,结果表明:融合后点云均方根误差为0.063 m,精度提高了0.018 m。
Deformation monitoring of high and steep slopes with UAV LiDAR technology assisted by oblique images
-
摘要: 针对高海拔峡域地形地貌环境下基于轻小型无人机载LiDAR对高陡边坡激光点云扫描数据缺失导致DEM重建及形变分析精度低的问题,优化设计了一种垂直于山脊线、变高飞行的无人机点云/多视影像数据采集,以及影像密集匹配点云辅助下LiDAR三维激光点云的滑坡群DEM重建方案,实现了复杂地形地貌下LiDAR点云数据安全、高效的采集,改善了高陡边坡DEM重建及形变监测的精度和完整性。该方法基于迭代最邻近点算法,将倾斜影像生成的点云数据与同期获取的LiDAR点云数据配准和融合,实现了LiDAR点云数据缺失补偿,进而构建出完整、高精度的DEM,并与往期倾斜影像生成的DEM进行差分,对三个典型滑坡体进行了高程形变分析。以青海龙羊峡水电站的高陡边坡滑坡群为研究区,利用实测的GNSS地面控制点进行实验验证,得出结论:融合后的LiDAR点云精度为0.063 m,比融合前提高了0.018 m;重建的三个典型滑坡体的DEM高程精度为0.08 m,提升了边坡DEM重建的完整性和精度;对三个典型滑坡体2018、2021年两期高程形变分析,表明:滑坡群中多个边坡发生不同程度的土体滑动,高程方向的形变高达50多米,滑坡群形变量大。Abstract: Aiming at the problem of low accuracy of DEM reconstruction and deformation analysis due to the lack of data during the scanning of high and steep slope point cloud based on light and small unmanned airborne LiDAR in the topographic environment of high-altitude gorge, this paper proposed an optimized data acquisition scheme using UAV image/point cloud data acquisition of high and steep slope flying perpendicular to the ridge, and DEM reconstruction scheme using UAV LiDAR point cloud with the aid of dense point cloud derived from multi-view oblique images. The accuracy and integrity of DEM reconstruction of high and steep slope and the accuracy of landslide deformation measurement in different periods have been improved. This method uses the point cloud data generated from the collected tilted image and the LiDAR point cloud data of the same period, in order to compensate the LiDAR point cloud missing data based on the iterative nearest neighbor point algorithm. Then, high-precision DEM products of landslides are constructed using the compensated LiDAR point cloud; Finally, the deformation of the landslides in elevation direction is obtained by comparing the current DEM with the DEM generated from the oblique images acquired in the past. Taking the landslides located at high and steep slopes of Qinghai Longyangxia Hydropower Station as the study area, an experiment is carried out and verified using the measured GNSS ground control points. It is concluded that the accuracy of LiDAR point cloud after fusion is 0.063 m, which is 0.018 m higher than that before fusion. The elevation accuracy of the reconstructed DEM of three typical landslides is 0.08 m, which improves the completeness and accuracy of DEM of slopes. The deformation results of typical landslides in the area are obtained by a comparison of DEM from two different epochs. The conclusions can be drawn that many slopes in the landslide group have different degrees of soil sliding, and the deformation in the elevation direction is up to more than 50 m, heavy deformations happened to the landslides.
-
Key words:
- oblique images /
- UAV LiDAR /
- point cloud registration /
- deformation monitoring /
- high and steep slopes
-
表 1 D-LiDAR2000激光雷达模块参数
Table 1. D-LiDAR2000 LiDAR equipment parameters
Ranging
modeWave-
length/
nmRanging
precision/
cmPoint
frequency/
kHzEcho
strength/
bitsHorizontal
accuracy/
mTOF 905 ±2 240 8 0.02 表 2 D-OP3000倾斜相机参数
Table 2. D-OP3000 oblique cameras parameters
Camera
selectionSize of
sensor/mmEffective
pixelsLens focal
length/mmSONYa6000 23.5×15.6 2.430 million×5 25(down)
35(oblique)注:×5指五个相机获取的实际像素 -
[1] Wang Qingguo, Zhao Hai, Li Jianping. Landslide monitoring by merging ground laser point cloud and aerial image [J]. Bulletin of Surveying and Mapping, 2019(4): 99-102. (in Chinese) doi: 10.13474/j.cnki.11-2246.2019.0122 [2] Chi Weikang, Li Xiangxin, Li Pengfei. Reservoir slope deformation monitoring based on GPS [J]. Journal of Geological Hazards and Environment Preservation, 2018, 29(1): 9-12. (in Chinese) [3] Wang Chunyang, Li Guorui, Liu Xuelian, et al. 3D point cloud registration algorithm with IVCCS [J]. Infrared and Laser Engineering, 2022, 51(5): 20210491. (in Chinese) doi: 10.3788/IRLA20210491 [4] Yuan Xiuxiao, Yuan Wei, Xu Shu, et al. Research development and prospect on dense image matching in photogrammetry [J]. Acta Geodaetica et Cartographica Sinica, 2019, 48(12): 1542-1550. (in Chinese) [5] Yao Yanli, Jiang Shengping, Wang Hongping. Landslide model monitoring and foreccast by using ground 3D laser scanner [J]. Science of Surveying and Mapping, 2014, 39(11): 42-46. (in Chinese) [6] Zhu Haixiong, Sui Lichun, Lu Kaixiang. Application of terrestrial 3D laser scanning technology in deformation monitoring of dangerous rock mass [J]. Bulletin of Surveying and Mapping, 2017, 11: 68-71. (in Chinese) [7] Yang Huishan, Lei Lizhen. Research on key technologies of acquisition of airborne LiDAR point cloud data and high-precision DEM [J]. Bulletin of Surveying and Mapping, 2019(3): 124-127. (in Chinese) doi: 10.13474/j.cnki.11-2246.2019.0092 [8] Zhao Fuqun, Zhou Mingquan. Hierarchical point cloud denoising algorithm [J]. Optics and Precision Engineering, 2020, 28(7): 1618-1625. (in Chinese) doi: 10.37188/OPE.20202807.1618 [9] Wang Cong. DEM production process based on airborne LiDAR data [J]. Geomatics & Spatial Information Technology, 2020, 43(10): 204-207. (in Chinese) doi: 10.3969/j.issn.1672-5867.2020.10.055 [10] Lv Yuanzhi, Sun Qiang, Bi Guoling. Recognition and repairing of surface hole in three dimensional laser scanning system [J]. Chinese Optics, 2016, 9(1): 114-121. (in Chinese) doi: 10.3788/co.20160901.0114 [11] Wang Xin, Zhang Mingming, Yu Xiao, et al. Point cloud registration based on improved iterative closest point method [J]. Optics and Precision Engineering, 2012, 20(9): 2068-2077. (in Chinese) doi: 10.3788/OPE.20122009.2068 [12] Xie Xiaopeng, Gu Jiawei. An improved two-dimension ICP point cloud registration algorithm [J]. Laser & Infrared, 2021, 51(7): 951-955. (in Chinese) doi: 10.3969/j.issn.1001-5078.2021.07.020 [13] Miao Qiguang, Shi Junjie, Liu Tiange, et al. New efficient DSM genertating algorithm based on TIN [J]. System Engineering and Electronics, 2014, 36(9): 1868-1873. (in Chinese) doi: 10.3969/j.issn.1001-506X.2014.09.32 [14] Chang Bingtao, Chen Chuanfa, Guo Jiaojiao, et al. Interpolation-based filtering with segmentation for airborne LiDAR point clouds [J]. Infrared and Laser Engineering, 2021, 50(9): 20200369. (in Chinese) doi: 10.3788/IRLA20200369 [15] Wang Yunkai, Jiang Yi, He Yanlan, et al. The application of airborne LiDAR and inclination photography in beautiful rural construction [J]. Bulletin of Surveying and Mapping, 2021(S1): 37-41. (in Chinese) doi: 10.13474/j.cnki.11-2246.2021.0508