摘要
SfM （Structure from Motion,运动重建结构） 是一种新的高精度三维形态模型数据获取技术,其应用于精细表面测量中的分辨率及精度尚待详细的评估.该文以化石表面为研究对象,重叠度70% 摄取化石表面的照片并生成高密度点云,同T LS（ T errestrial Laser Scanner,地面三维激光扫描仪） 点云进行精配准后生成 DEM .以TLS 点云及D E M 作为比较对象,经过化石表面微小特征与剖面解析、点云、DEM 偏差计算和误差分布的分析和验证,结果显示：1） SfMDEM 实际表面水平及垂直分辨率分别可达到1 mm 和2 mm .2） S fM 测量精度非常接近T LS, SfM点云的R MSE 为1. 3 mm ,平均垂直偏差为1. 5 mm; SfMDEM 的R MSE 为1 9 mm ,平均绝对高程偏差为2 2mm .3） SfM 测量误差在大于55〉的坡度范围内与坡度呈正相关,在阴影值0-170 区间内与阴影呈反相关,照片拍摄角度不完整、数量不充分的区域误差较大.研究为SfM 测量的精细程度提供了参考,也为地学应用中表面微小节理识别及剖面分析提供了便利可靠方法.
As a new technology to acquire high-resolution Digital Elevation Models(DEMs),the resolution and accuracy of SfM(Structure from Motion)measurement for fine surface survey remains unclear.In this study,we took 111 parallel photos with70% overlapping on a fossil surface,and generated high density SfM point cloud by Agisoft Photoscan software.To compare with the SfM measurement,a single station TLS measurement was also carried out,and the point cloud was acquired.The SfM point cloud was then calibrated by the TLS point cloud and 1mm SfM DEM was constructed through IDW interpolation.Comparisons between the SfM and TLS data indicate that:1)The resolution of SfM DEM is better than TLS DEM,and the horizontal and vertical resolution of SfM DEM can achieve 2mm and 1mm,respectively.2)The accuracy of the SfM data is comparable to that of the TLS data.For the SfM point cloud data,the Root Mean Square Error(RMSE)at the calibration by the TLS point cloud is 1.3mm,and the mean vertical deviation is 1.5mm.For the 1mm SfM DEM,the RMSE and the mean height deviation are 1.9mm and 2.2mm,respectively.3)The error of SfM measurement is effected by slope angle,shade and shooting angle.The larger error appears in the area with steep slope or in shadows.In addition,insufficient of shooting density or shooting angle can bring larger error too.The DEM error shows positive relation to the slope degree in range of 55°to 90°,and shows negative relation to the hill-shade value in range of 0to 170.The results are helpful for better understanding and further application of SfM measurement in geologic and geographic studies.
出处
《地理与地理信息科学》
CSCD
北大核心
2016年第6期25-31,共7页
Geography and Geo-Information Science
基金
国家社会科学基金项目(13&ZD192)
国家自然科学基金项目(41472174)
浙江省文物保护科技项目(2014012)
