3-D reconstruction of underwater objects using image sequences from optical camera and imaging sonar

Abstract

This paper proposes a method for 3-D reconstruction of underwater objects by using an optical camera and an imaging sonar. The optical camera built a volumetric model by performing simultaneous localization and mapping, whereas the imaging sonar built another volumetric model by using a space-carving approach. Two volumetric models corrected each other, and then the corrected models were integrated into a hybrid model to complete the reconstruction. The proposed method could reconstruct underwater objects with less volumetric errors than conventional reconstruction methods, but required more computation time. The proposed method could be applicable to underwater missions where the reconstruction accuracy is more valuable than the operation time. © 2019 Marine Technology Society.