典型高原坡耕地的UAV重复观测3D检测精度实证分析  

An Empirical Analysis of 3D Detection Accuracy of UAV Repeated Observation for the Typical Slope Farmland of Dongchuan Red Land

作　　者：高莎 甘淑[1,2] 袁希平 胡琳[1] 毕瑞 李绕波 罗为东 GAO Sha;GAN Shu;YUAN Xi-ping;HU Lin;BI Rui;LI Rao-bo;LUO Wei-dong(School of Land and Resources Engineering,Kunming University of Science and Technology,Kunming 650093,China;Research Center of Applied Engineering of Spatial Information Surveying and Mapping Technology in Plateau and Mountainous in Yunnan Province,Kunming 650093,China;Key Laboratory of Mountain Land Cloud Data Processing and Application for Universities in Yunnan Province,West Yunnan University of Applied Sciences,Dali 671006,China)

机构地区：[1]昆明理工大学国土资源工程学院,云南昆明650093 [2]云南省高校高原山地空间信息测绘技术应用工程研究中心,云南昆明650093 [3]滇西应用技术大学云南省高校山地实景点云数据处理及应用重点实验室,云南大理671006

出　　处：《光谱学与光谱分析》2023年第7期2188-2194,共7页Spectroscopy and Spectral Analysis

基　　金：国家自然科学基金项目(41861054)资助。

摘　　要：随着低空无人机(unmaned aerial vehicle,UAV)技术的快速发展,搭载光学传感器的小型消费级无人机可快速、灵活地获取目标对象的高分辨率影像数据,在地学各领域中呈现出广阔的应用前景。UAV-SfM(structure from motion)作为成像3D立体构建的最新技术方法,是深化低空无人机技术地学领域研究的核心技术,但目前对于运用UAV-SfM方法获取数据综合精度的研究不足,影响了该技术进一步的推广应用。针对DJI Phantom 4 RTK消费级无人机是否具有应用于滇中高原山地浅层地表变化检测的可能性问题,选择了东川红土地典型坡耕地为试验区,采用相同航高重复飞行规划方案,并运用SfM-MVS关键技术处理获取同一测区重复的DSM与DOM数据。为了评价分析针对典型坡耕地的UAV重复观测的测量精度,特别对实验区中的裸露坡耕地Ⅰ和有作物生长的坡耕地Ⅱ,分别采用基于剖面线的3D离散点抽样和基于窗口面的3D点集抽样方法,开展了对UAV重复观测坡耕地的3D点位精度评价。点位精度分析表明:①基于剖面线的3D离散点抽样及精度分析,坡耕地Ⅰ平面点位精度误差均值为±0.029 m,3D点位误差精度为±0.072 m;坡耕地Ⅱ平面点位精度误差均值为±0.032 m,3D点位误差精度为±0.075 m。②基于窗口面的3D点集抽样及精度分析,坡耕地Ⅰ平面点位精度误差均值为±0.013 m,3D点位误差精度为±0.066 m;坡耕地Ⅱ平面点位精度误差均值为±0.038 m,3D点位误差精度为±0.076 m。综合分析得出,基于剖面线单点抽样评价精度结果整体好于基于窗口的3D点集抽样评价精度,但总体上平面精度与垂直精度均能够达到厘米级。实验对比分析研究得出,不同地表粗糙度对UAV重复观测精度存在影响,地表粗糙度大的3D点位误差比粗糙度小的点位误差大。该研究成果可以为基于无人机与SfM方法的地貌数据采集与三维重建的精度控制和采集方案设With the rapid development of low-altitude unmanned aerial vehicle technology,small consumer-grade UAVs equipped with optical sensors can quickly and flexibly acquire high-resolution image data of target objects,which presents a broad prospect for expanding applications in various fields of geology.UAV-SfM is the latest technological method for imaging 3D stereo construction that is a core technology for deepening research in the field of low-altitude UAV technology geology,but at present the lack of research on the comprehensive accuracy of data results obtained by using the UAV-SfM method has affected the further promotion and application of this technical method.This paper addresses whether the DJI Phantom 4 RTK consumer-grade UAV has the technical potential to be used for detecting shallow surface changes in the mountainous areas of the central Yunnan plateau,and selects a typical sloping field in the red land of Dongchuan as the test area.The DSM and DOM data from the same area were obtained using the key SfM-MVS technique.In order to evaluate the accuracy of the repeated UAV observations on typical sloping land,3D point accuracy evaluation was carried out for the bare sloping land I and the growing sloping landⅡin the experimental area,using 3D discrete point sampling based on profile lines and 3D point set sampling based on window surfaces,respectively.The point accuracy analysis showed that:(1)In 3D discrete point sampling and precision analysis based on profile line,the mean precision error of slope farmlandⅠplane point position is±0.029 m,and the accuracy of 3D point position error is±0.072 m.The mean accuracy of slope farmlandⅡplane point position is±0.032 m,and the 3D point position error is±0.075 m.(2)Based on the 3D point set sampling and precision analysis of the window surface,the mean precision error of slope farmland I plane point position is±0.013 m,and the 3D point position error accuracy is±0.066 m.The mean precision error of the slope farmlandⅡplane point position is±0.038 m,a

关 键 词：UAV SfM-MVS DSM 点位误差 精度分析 

分 类 号：P237[天文地球—摄影测量与遥感]