The potential for creating a geometric digital twin of road surfaces using photogrammetry and computer vision

Abstract: Current methods to create 3D models of roads are not scalable. Advances in photogrammetry mean that they have emerged as a realistic alternative to LiDAR for producing 3D spatial data. As photogrammetry is based on image data, methods developed in the area of computer vision can be utilised to segment or find relationships between assets in the point cloud. This paper presents the various advances in computer vision that can be applied to generate 3D models of the road surface and surface assets. This could in turn be used as a basis for creating a digital twin of a road.