3D reconstruction of shallow sea structures using direct system calibration and faint laser line extraction.

Virtually recreating an underwater scene requires 3D data of its objects. Creating a dense 3D point cloud of shallow underwater objects using a laser in the presence of significant ambient light is challenging. Although several methodologies have been proposed for underwater reconstruction, they essentially work in low-light environments. In particular, current techniques are unable to quickly and effortlessly generate dense point clouds of underwater objects under varying light. The present work fills that gap. The system uses a rotating low-power line-laser and is designed in such a way that it works in the dark as well as in sunlight, which tends to wash out low-power laser in shallow waters. In order to achieve this goal, we propose an analytical technique to segment faint-laser lines in an underwater lit environment, which does not depend on the availability of large training data. Thereafter, the extracted laser lines are applied for triangulation to generate the point cloud online and automatically. To reduce the cumulative errors during calibration, our data-centric method calibrates the entire system directly, unlike existing methods, where multiple camera and system parameters are calibrated. An evaluation of the performance of the proposed method in different lighting conditions is reported.