Seabed Terrain 3D Reconstruction Using 2D Forward-Looking Sonar: A Sea-Trial Report From The Pipeline Burying Project

摘要

In this short paper, we report some preliminary results on the seabed terrain reconstruction using the two-dimensional (2D) forward-looking sonar. In the underwater pipeline burying project, when the burying machine excavating the trench using the jetting device, the stirred turbid water sharply shorten the visual distance of optical camera. Researchers from Shenyang Institute of Automation, Chinese Academy of Sciences (SIA, CAS) and Korea Institute of Robot and Convergence (KIRO) work together to reconstruct the 3D environment using the acoustic camera. The proposed reconstruction procedure mainly includes two parts: 3D point-cloud extraction for each sonar frame and point-cloud registration. With the former step, the elevation information, which is irreversely lost in the acoustic imaging process, is recovered by the shape-from-shading technology under the Lambert surface assumption. With the latter, a registration method based on symmetrical Kullback-Leibler divergence has been proposed to stitch the point clouds from each sonar image into a panoramic terrain map. The initial experiment results with the sea-trial datasets demonstrate that the proposed method is feasible for the practical underwater engineering.