UAV-based automatic generation of high-resolution panorama at a construction site with a focus on preprocessing for image stitching

摘要

As construction sites increase in size, it becomes more difficult for a manager to understand the status of the site on time. However, with the development of unmanned aerial vehicles (UAVs), it is possible to collect a large amount of visual data of the construction site in a short time. Using this data, a large-scale construction site can be monitored in a timely and frequent manner with computer vision technologies. This paper proposes a method to generate a panorama of a construction site by using an image stitching technique with a focus on preprocessing. To create high-quality panoramas, blurred frames of videos are filtered out, key frames are selected, and camera lens distortion is corrected. The proposed method produced a high-quality panorama of a construction site, which was evaluated by comparing it with an aerial photograph and the panorama produced by the existing image stitching technique. The proposed method is expected to help managers to easily identify various construction site conditions with the help of high-quality image data.