Automatic Mosaic Method of UAV Water-Area Images Based on POS Data

This paper mainly studies the automatic mosaic method of Unmanned Aerial Vehicle (UAV) large water-area images. During the mosaic process, the monitoring water area was relatively larger and the image gradation changed small, both of which caused little effective feature points and can be detected. In order to solve this problem, Scale Invariant Feature Transform (SIFT) and Harris algorithms were synergistic used to extract the feature points, when there was not only water area but also one small land in images. Meanwhile, POS directional data was used to help the geometric correction of images in order to achieve image mosaic when there was only water area. Finally, this paper takes oil spills monitoring of offshore surface as experimental object to describe the UAV practical application value in terms of water resources monitoring and emergency relief.