Marine Debris Monitoring on Uninhabited Islands Using UAV-Based 3D Modeling

Abstract. Uninhabited islands in South Korea play a pivotal role in maritime activities and are increasingly vulnerable to the threat of marine debris. This paper proposes a UAV (unmanned aerial vehicle)-based 3D analysis to estimate the volume of marine debris. The proposed method consists of three main steps: bundle adjustment, 3D reconstruction, and volume calculation. Tiepoints are extracted from UAV images and used for bundle adjustment to correct the EOPs (exterior orientation parameters) and generate a sparse point cloud. The sparse point cloud is then densified through stereo matching, utilized for 3D mesh model generation. The volume of marine debris is finally calculated based on tetrahedral decomposition from the 3D mesh model. Three uninhabited islands in South Korea were selected as experimental areas. Due to the inaccessibility of the islands, a total of less than 30 images per site were acquired. Nevertheless, the coastal terrain and debris features were successfully reconstructed. The proposed method estimated the volume of a single Styrofoam buoy with a 0.0026 m³ error. Furthermore, it can estimate the volume of clustered debris without assuming a specific height. The findings demonstrated that the proposed method could overcome the limitations of 2D analysis and provide precise volume estimations. This method is expected to be effective for remote monitoring of marine debris on uninhabited islands.