| Abstrakt: | In this thesis we focus on the problem of automatic scanning of a general 3D world object or a scene using a drone. We mainly address the flight mission planning part, where the locations and look-at directions of the drone`s camera are calculated as well as the path planing. New camera positions and a path, that visits all the camera positions with a drone, are calculated from the sparse mesh and camera poses calculated with photogrammetry. We find regions in the sparse model with insufficient detail. These regions are then clustered into bigger regions in a way, that they can be photographed in whole. Camera positions, which are viewing the regions with small detail are constructed to have a clear line of sight to the given region. The routed flight, that visits all the camera positions is evaluated to be unobstructed and to have the smallest possible flight time. This process of flight and processing can be repeated until we get our desired reconstruction.
|
|---|