Adaptation analysis of produced 3D models from UAV-SLAM and UAV-TLS data combinations

Abstract

Photogrammetric techniques are widely used to represent the characteristics of historical buildings in the digital environment in the closest way to reality. Terrestrial photogrammetric methods have needed different alternatives in recent years to produce 3D models that offer high reality. In recent years, producing point data with the Terrestrial Laser Scanner (TLS), Unmanned Aerial Vehicles (UAV) images, and alternative methods such as the Simultaneous Localization and Mapping (SLAM) algorithm has become popular. Aligning point data from different approaches and making 3D models create new problems. Primarily, it is necessary to investigate the operations performed manually by the operator over time. Also, it is needed to explore the integration of automatic algorithms such as Iterative Closest Point (ICP) in terms of accuracy. In this research, point data of the Dokuz Historical Bridge was produced using different aligning techniques manually and automatically with the ICP algorithm. The assessment has been done from the results of combined point data over time and accuracy. In both UAV-TLS and UAV-SLAM aligning procedures, spatial accuracy was determined as 2.8cm and 4cm, respectively, in the operations performed by the operator and 46cm and 12cm in the procedures performed automatically by ICP. As a result, it was determined that the operators combined produced better findings, especially in research including coordinate transform.