Enhancement of digital surface model in dense urban regions obtained from UAV photogrammetry in flat and steep terrains

Abstract

Generating accurate 3D models of the city or a region has become a key important step to plan any further development or modifications from the current plan. Close-range remote sensing technologies have been very efficient in generating accurate 3D models of the regions with very detailed information. Unmanned Aerial Vehicle (UAV) is one such technology that is widely used due to their cost-effectiveness, time efficiency, and for various other reasons. In this research, the quality of the Digital Surface Model (DSM) produced through the photogrammetric software process is tried to enhance in some parts of the datasets. The DSM is generated from the pointcloud generated from Structure from Motion (SfM). In dense urban regions, the overlap of the UAV images is not sufficient or incapable to generate ample points in the point cloud at some places in the dataset, due to which the DSM generated is not that accurate. The point cloud is enhanced using an interpolation technique at those regions. From which the DSM generated is enhanced. Qualitative analysis is also done between the DSM obtained through the photogrammetric software process and the DSM enhanced through the proposed methodology. This low-cost approach can be implemented while creating digital twin models.