SfM摄影测量法测定切沟的精度评价
史扬子1, 黄婷婷1, 罗建勇1, 杨 扬1, 刘宝元2

(1.北京师范大学 地理科学学部 地表过程与资源生态国家重点实验室, 北京100875; 2.中国科学院 水利部 水土保持研究所 黄土高原土壤侵蚀与旱地农业国家重点实验室, 陕西 杨凌 712100)

SfM摄影测量法; 切沟测量; 实时动态差分GPS; 地面控制点; 数字高程模型

Evaluation on the Accuracy of Structure From Motion（SfM）Photogrammetry on Gully Surveying
SHI Yangzi1, HUANG Tingting1, LUO Jianyong1, YANG Yang1, LIU Baoyuan2

(1.State Key Laboratory of Earth Surface Processes and Resource Ecology, Faculty of Geographical Science, Beijing Normal University, Beijing 100875, China; 2.State Key Laboratory of Soil Erosion and Dryland Farming on the Loess Plateau, Institute of Soil and Water Conservation, Chinese Academy of Sciences and ministry of Water Resources, Yangling, Shaanxi 712100, China)

structure from motion(SfM)photogrammetry; gully survey; real-time kinematic(RTK)GPS; ground control point; digital elevation model(DEM)

备注

选取河北省滦平县两间房村的一条典型切沟,以实时动态差分(Real-time Kinematic,RTK)GPS的测量结果为基准,评估了断面测量法与SfM方法对该切沟的测量精度,并在此基础上探讨不同地面控制点(Ground Control Point,GCP)布设方案对SfM测量精度的影响。结果 表明:SfM测量法可作为一种高精度的方法运用于切沟测量,与RTK GPS相比,断面测量法的体积测量误差高达37.28%; 而SfM方法的误差仅为2.40%,且相对RTK GPS所测切沟DEM的差值主要集中于-0.2～0.2 m。此外,轻便、小巧的激光测距仪可有效替代RTK GPS用于GCP坐标的测量,在此基础上得到的切沟体积误差仅为0.23%。将切沟沟缘与沟底共18个GCP抽稀至4～16个,并利用SfM方法重建该切沟的三维形态。各方案所得切沟DEM相比未抽稀结果的差值主要集中于-0.05～0.05 m,且体积差值百分比均低于5%。研究结果证实,SfM摄影测量法在切沟测量中具有较高的精度,可应用于野外切沟的快速、精确测量,对于切沟定量监测与切沟侵蚀规律研究具有重要的实践意义。

A typical gully was selected and measured using the tape method and structure from motion(SfM)photogrammetry in the village of Liangjianfang, Luanping County, Hebei Province. The accuracies of both gully surveying methods were evaluated via comparing the results with those obtained by the real-time kinematic(RTK)GPS. In addition, different spatial distribution schemes of ground control points(GCPs)used in the SfM photogrammetry were designed and their impact on the gully measurement accuracy was assessed. The results show that SfM photogrammetry can serve as a highly accurate technique for gully surveying. Compared to RTK GPS, the tape method misestimates the gully volume by 37.28%, whereas the SfM photogrammetry overestimates the volume by only 2.40%, when the DEM obtained by RTK GPS is subtracted from the one by SfM photogrammetry, the resulting elevation difference falls between -0.2 m and 0.2 m for more than 70% of the pixels; moreover, the laser range finder(LRF)was demonstrated as an effective alternative for the determination of GCP coordinates; for each pair of GCPs, LRF generates statistically similar spatial distance and elevation difference with the classical RTK GPS; furthermore, the gully volume derived by SfM photogrammetry based on such GCP measurement is only misestimated by 0.23% compared to the volume obtained by RTK GPS. Reducing the number of GCPs on the gully edge and bottom from 18 to a range between 4 and 16, little difference is detected in the corresponding DEMs reconstructed by the SfM photogrammetry. Compared to the DEM grounded on all the 18 GCPs, the elevation difference is within the range between -0.05 and 0.05 m for most pixels, disregarding the number and locations of the GCPs considered. The estimation errors for gully volume, accordingly, are all smaller than 5%. These findings demonstrate that SfM photogrammetry has high precision in terms of gully measurement, and is applicable for rapid and accurate gully survey in the field, which has great practical implications for gully monitoring and gully erosion research.