Underwater terrain matching method based on multibeam bathymetric point cloud descriptors

• Published in: International journal of digital earth Vol. 17; no. 1
• Main Authors: Long, Jiawei, Zhao, Jianhu, Zhao, Xi, Jin, Chengxi
• Format: Journal Article
• Language: English
• Published: Abingdon Taylor & Francis 31.12.2024; Taylor & Francis Ltd; Taylor & Francis Group
• Subjects: acoustics; Bathymetric data; data collection; Doppler sonar; Global navigation satellite system; global positioning systems; Inertial coordinates; Inertial navigation; Inertial platforms; landscapes; Localization; Matching; Multibeam echosounder sounding system; Navigation; Navigational satellites; Neural networks; Ocean floor; point cloud descriptor; Positioning systems; Similarity; Terrain; Underwater; underwater navigation; underwater terrain matching
• ISSN: 1753-8947; 1753-8955; 1753-8955
• DOI: 10.1080/17538947.2024.2434655

In the absence of global navigation satellite system (GNSS) signals and acoustic positioning systems, and relying only on inertial measurement unit (IMU) and Doppler velocity log (DVL), underwater terrain matching has become the primary approach of underwater navigation and localization. To address the limitations of current underwater terrain matching methods, which heavily depend on high-precision background fields of seafloor terrain and are subject to the richness of seafloor terrain information, we propose a novel underwater terrain matching method based on multibeam bathymetric point cloud descriptors. This method generates discriminant descriptors from the bathymetric point cloud patches, which can be directly used to accurately measure the similarity between two patches to complete the matching. This approach eliminates the need for recalculating similarity between different patches and reduces memory requirements for storing original bathymetric data. Specifically, our method fully considers the principle of multibeam data measurement and includes a patch construction method of multibeam bathymetric point cloud and a terrain descriptor generation model based on point cloud neural networks. We compared the proposed method with other state-of-the-art underwater terrain matching methods on both a test set and real-world data. The results demonstrate that our method exhibits superior matching performance.