A REAL-TIME DRONE MAPPING PLATFORM FOR MARINE SURVEILLANCE

Abstract. Marine incidents have caused serious casualties and damaged on property, and situational awareness and actions are needed to reduce further extensive damage for marine surveillance. The importance of an attempt for maritime monitoring using UAV has been raised, and a platform should be prepared to respond immediately to urgent situations. In this research, a real-time drone image mapping platform is proposed for marine surveillance that receives marine images acquired and transmitted by drones and processes them in real time. The platform proposed in this study is divided into 1) UAV System, 2) Real-time image processing, 3) Visualization. UAV system transfers data from a drone to the ground stations. Real-time image processing module generates individual orthophotos followed by directly georeferencing in real time and detecting ships on the orthophotos. Visualization module enables to visualize the orthophotos. The overall mapping time of 3.26 seconds on average was verified for processing image mapping, and ship detection time for a single image was estimated to be within about 1 second, which corresponds to an environment in which an emergency must be handled. In conclusion, a real-time drone mapping platform that is introduced in this study can be evaluated as being available for maritime monitoring that requires swift responses.