An Automated Method for 3D Modeling of Underground Pipe using RANSAC

Abstract

Underground pipes, such as gas and water pipes, are essential components of social infrastructure. However, since these pipes are located underground, it is difficult to accurately determine their position and shape. On the other hand, in recent years, it has become easy to capture photographs using smartphones and digital cameras, enabling the generation of TIN meshes using techniques such as Structure from Motion (SfM). By utilizing these sensors, there is potential to efficiently acquire 3D information. In this paper, we discuss a method for automatically creating 3D surface models from TIN mesh using the RANSAC algorithm. An exhaustive search of all possible parameter combinations was performed to determine the optimal parameters. Furthermore, evaluation experiments using real-world data confirmed that the method can appropriately model underground pipes.