Institute of Transportation, MOTC

       Unmanned Aerial Vehicle (UAV) technologies have shown dramatic development in recent years. Capable of carrying highresolution cameras and with the function of high maneuverability and remote control, UAV can replace human labor and go safely and fast where it was inaccessible. Currently by combining with image processing, UAV technologies have further been widely applied to data acquisition tasks in large areas or areas of particular environmental conditions as an ideal option for long-term monitoring to keep the safety of the facilities in the harbor where human resources are limited. This project aims to build a web-based tracking platform with UAV and image analysis techniques. The study was intended to be implemented in two years. In the first year, a high-resolution 3D orthophoto map of the 4,700-ha port area was generated. That is followed by making the software process for UAV flight route planning by integrating UAV control and image acquisition logic; the details of robust images captured by that process will be put to image feature matching. Finally, a platform for system integration will be implemented. The second year followed the results of the first year, some improvements were done. Firstly, three correction methods were utilized to improve the performance of image matching. Furthermore, the UAV can successfully fly along the planned path, fly stably with the vehicle, and regularly collect images for further analysis. In terms of object identification, we presented the result in a visual way. Meanwhile, the PCI index was proposed to evaluate the road surface and the results were also verified successfully. Those methods and results can be applied in the patrol tasks of the port, tracking the condition of land use and the security of certain facilities.