Condition assessment of ship structure using robot assisted 3D-reconstruction

摘要

Ships condition is assessed regularly to maintain safety. Traditionally, structural integrity assessment is performed by surveyors, requiring complex and time-consuming operations to accessany ship space. Imagery based, three-dimensional (3D) reconstruction of structures is a new area obtaining considerable interest. It can provide low-cost, less disruptive and safer inspection approach. In this study, alternative technologies to generate 3D models, based on photos, are explored. The aim is to highlight how human made ship survey can be improved using robotics technology. A procedure for 3D reconstruction combining photogrammetry/videogrammetry and computer vision techniques is developed, providing an alternative to ease vessel inspections. Moreover, effect of pre-processing image datasets, aimed at improving the performance of 3D reconstruction, is investigated. An efficient image pre-processing pipeline based on computer vision algorithms for colour enhancement, shadow removal and image blurriness is presented. This study can help in effective and reliable decision-making process, due to its user-friendliness and cost effectiveness, mainly for cargo holds requiring frequent assessment because of cargo operations induced damages.