A Case Study on the Accuracy Assessment of a Small UAS Photogrammetric Survey Using Terrestrial Laser Scanning

摘要

Small unmanned aerial system (sUAS) technology is rapidly transforming the manner in which geospatial data are collected in the surveying and mapping industry. Topographic datasets, which in the past have been collected with manned aircraft and aerial photogrammetry or time-consumingcollection techniques on the ground, can now be compiled more quickly and at a lower cost with the use of sUAS. However, sUAS surveys present limitations, as the accuracy of derived digital elevation models (DEM) depends on several factors (e. g., camera type, camera calibration, shadows,and altitude). The objective of this study was to assess the accuracy of DEMs that are derived from sUAS surveys and analyze the spatial distribution of DEM errors. In this study, data originate from a sUAS survey, conducted on a portion of the Pennsylvania State University, Wilkes-Barre campus.Accuracy assessment of derived DEMs is accomplished through comparison with collocated terrestrial laser scanner point clouds. The resulting statistical parameters show that the size of DEM errors depends on scene complexity, as considerably lower accuracies are obtained in complex surfacessuch as roofs and domes (e. g., root-mean-squared error (RMSE) of few centimeters to decimeter on complex surfaces versus 1-2 cm in noncomplex areas such as parking lots). In addition, histograms of height comparisons in noncomplex areas present better agreement with the normal distributionthan histograms in noncomplex areas. Furthermore, this study investigates the number of checkpoints needed for reliable accuracy assessment of DEMs. Results show that about 60 and 150 checkpoints are required to maintain a deviation of ±20 percent and ±10 percent from the referenceRMSE value, respectively. Thus, this study enhances our understanding of the spatial variability of DEM errors, and aids in accuracy assessment procedures.