Simulated surface anomaly detection in underwater object reconstruction

This paper addresses simulated sonar-based underwater inspection mission (UIM) and automated surface anomaly detection of underwater machinery. In order to ensure harbor safety, foreign ships are inspected prior entering harbor, with main goal of automated mine like objects detection. UIM is carried out by an autonomous underwater vehicle (AUV). Trajectory of a single AUV is planned specifically for a particular ship hull. As a result of UIM, a 3D point cloud is reconstructed and it’s orthographic projection is used for creating 2D image patches. Subsequently, image processing methods are applied in order to automate surface anomaly detection. Computer models of harbor wall and ship’s hull are considered in this paper. Mine- like object insertion and detection methods are proposed and evaluated.