Generation of 3D surface models from UAV imagery varying flight patterns and processing parameters

The potential of Unmanned Aerial Vehicles (UAVs) as a low-cost survey technology platform has been widely demonstrated in several research studies. The active development of more refined survey techniques and Computer Vision (CV) algorithms makes these platforms effective and efficient in terms of results and performances. On the other hand, however, products quality is strictly influenced by flight planning of each survey and the processing parameters of imagery. In this work, a comparative analysis of different 3D models derived from UAV images was carried out. Several image datasets acquired with different percentages of images overlap, or subjected to different down-sampling, were acquired and processed, evaluating the contribution of the camera pre-calibration and the influence of each image metadata in dataset processing.