High-resolution underwater mapping in low-visibility and confined environments using imaging sonar

The detailed characterization of low-visibility and confined underwater environments remains a critical challenge for autonomous exploration due to the limited spatial resolution of conventional sensing techniques. Previous mapping methods often depend on complex multi-sensor systems, which increase operational cost and reduce adaptability across diverse environments. This study proposes a novel high-resolution mapping approach using a single imaging sonar, which integrates acoustic mosaicking techniques with a shadow cue-based target height estimation model. This approach can be regarded as a 2.5-D mapping strategy: a planar mapping is first generated from 2-D sonar images, and the estimated target heights are then used to compensate for the missing elevation information to achieve a spatial representation. Unlike conventional 3-D reconstruction techniques, it emphasizes the rapid acquisition of spatial contours rather than the recovery of structural details. Experiments conducted in controlled pool environments demonstrate that the proposed method can reliably reconstruct spatial layouts and target dimensions, offering a robust and scalable solution for autonomous underwater perception. The approach advances low-altitude surveys in extreme environments, opening new pathways for efficient perception using marine robotic systems.