ORIENTATION OF UAV IMAGE BLOCKS BY SURFACE MATCHING
- 1University of Porto, Science Faculty, Rua Campo Alegre, 416-007 Porto, Portugal
- 2Academia Militar, R. Gomes Freire 203, 1169-203 Lisboa, Portugal
Keywords: UAV, 3D Point Cloud, Aerial triangulation, Exterior orientation, Surface matching
Abstract. Topographic maps are produced in Portugal by the Army Geospatial Data Centre, based on a geographical database collected from aerial photography in resolutions from 0.30 to 0.50 m. Each map sheet is revised with an update interval of 10 years or more. Many changes, such as new roads or power lines, would be possible to update with much higher frequency using UAV imagery. Although the nominal scale of the published paper map sheets is 1 : 25,000, the actual positional accuracy requirement is of 1 meter, which is compatible with larger scales. Exterior orientation parameters obtained by UAV navigation equipment does not have enough accuracy for that. This paper deals with a method to automatically improve the UAV exterior orientation, intended to be compatible with the positional accuracy standards of the geographic database. The method is based on matching a point cloud derived from UAV imagery, without GCPs, with a reference DSM obtained from conventional aerial photographs, which were oriented by standard aerial triangulation. Results allowed to improve the UAV data to a positional accuracy of 1 meter, making it compatible with the requirements established for the geographic database.