Aerial Triangulation Using Different Time Images of Urban Areas Obtained from Unmanned Aerial Systems

Abstract. In case to ensure the development of urban areas, it is necessary to create and promptly update spatial data, which are primarily created using aerial photography. With the development of unmanned aerial systems, the survey of cities is mainly carried out with their help. However, due to the large amount of interference or failures in the operation of GNSS equipment installed on the UAV, the aerial photographs miss the coordinates of the projection centers or their quality does not allow to accurately process the photos for further creation of spatial data. It leads to perform extra field work, which affect time and cost of production.
To solve this problem, we propose the technology of using archival aerial survey data as a source for geodetic reference in the alignment of the new aerial images for same areas, which allows you to reduce the time and cost of work. Therefore, the purpose of this work is to investigate the possibility of the aerial triangulation using different time images of urban areas obtained from UAVs.
The research shows the results of experimental study on UAVs survey of two urban areas and includes flights made in 2018–2024 years with GSD 5–7 cm. In order to find tie points, we limit the searching area for key points using masks generated with the neural network YoloV8_p6. The accuracy of the resulted alignment was controlled using ground-based check points.
The performed research allowed us to confirm the validity of the proposed technology for using archival aerial photos to geo-reference new ones not only for aerial survey from manned aircrafts but also for photos from UAVs. The technology of aerial triangulation using different time images of urban areas can be recommended for aerial survey from manned vehicles for cities with population of millions of people, as well as aerial survey from unmanned vehicles when updating less populated city's spatial data.