Comparative Accuracy Assessment of two Low-Cost Devices for Underwater Structure-from-Motion 3D Reconstruction

Abstract. The accurate three-dimensional documentation of underwater environments is critical for infrastructure integrity assessments and archaeological studies, yet it often relies on costly equipment. Advances in low-cost Structure-from-Motion techniques provide accessible alternatives to conventional methods, enabling detailed reconstructions using consumer-grade devices. This study evaluates and compares the 3D reconstruction accuracy of two low-cost imaging systems, GoPro HERO13 and Huawei P30, in underwater and dry conditions, using three test objects with varying surface characteristics. High-precision reference models were generated via an ATOS Triple Scan system to quantify absolute accuracy. Results highlight significant performance differences between the cameras in dry conditions, with the GoPro showing improved reliability in underwater conditions, albeit facing challenges with lighting and low contrast environments. However, the Huawei P30 offered superior image resolution but suffered from usability issues in aquatic settings. Overall, dry condition scans yielded higher-quality point clouds, while underwater reconstructions were affected by environmental factors such as reflections, contrast, and lighting limitations. These findings provide practical insights into optimizing workflows for low-budget underwater photogrammetry and suggest potential improvements, including enhanced lighting setups and alternative imaging devices for future studies.