Comparative Analysis of UAV Photogrammetry and Terrestrial Laser Scanning for 3D Building Reconstruction

Abstract. Recently, unmanned aerial vehicle (UAV) photogrammetry has gained popularity as an alternative to terrestrial laser scanners (TLS) for collecting 3D information, particularly for building models. However, comprehensive comparative assessments of the two methods remain limited in the literature. This study compares TLS and UAV-based photogrammetry for 3D building modeling. 3D point cloud data is collected in a study area using both methods, with ground truth data collected using a total station. The resulting point clouds were evaluated through multiple quantitative metrics, including RMSE, cloud-to-cloud distance, surface area, point density, planarity, roughness, and surface variation, in addition to qualitative assessments of model completeness and feature reconstruction. Results show that TLS achieved lower RMSE and smoother, more accurate facade geometry, particularly in near-ground areas. However, this method failed to capture rooftop structures due to occlusion. The UAV model successfully captured upper structures and finer architectural features but showed noisier surfaces and missing data at the building base. Surface geometry analyses further revealed that TLS outputs were more planar and consistent. Meanwhile, UAV data exhibited greater variation and reconstruction artifacts. The findings highlight the strengths and limitations of each method depending on the modeling objective. While UAV photogrammetry may be sufficient for applications such as solar potential estimation or volume analysis, TLS is more suitable for high-precision tasks like facade documentation or structural monitoring. For projects requiring both accuracy and coverage, a hybrid approach is recommended. This study emphasizes their complementary value and offers guidance for urban-scale 3D data acquisition strategies.