Underwater topographic mapping using a fusion of Airborne LiDAR bathymetry and multibeam Echo sounding

Accurate measurement of underwater topography is of significant importance for maritime affairs, marine resource development, and environmental protection. Airborne LiDAR Bathymetry (ALB) can efficiently acquire integrated above-water and underwater data in shallow water areas, while the Multibeam Echo-sounding System (MBES) can precisely measure underwater topography in deep water areas and some shallow water regions. By integrating underwater acoustic and optical data, an integrated above-water and underwater topographic representation can be achieved. However, due to differences in measurement ranges, point cloud accuracy, and density, existing fusion algorithms perform poorly in combining ALB and MBES point clouds. This paper proposes a point cloud fusion method suitable for ALB and MBES data. The proposed method first uses a virtual grid combined with the alpha shape algorithm to quickly determine the overlapping area and removes edge triangles using boundary points, reducing the influence of edge triangles. Then, to minimize the impact of topographic differences on coarse point cloud registration, a point-to-plane ICP (Iterative Closest Point) algorithm based on TIN (Triangulated Irregular Network) grid vertex cutting planes is applied, with weighting assigned to the points corresponding to the matching cutting plane’s grid vertices. To mitigate the impact of point cloud density, the distance loss term in the Non-Rigid ICP cost function is changed from a point-to-point distance loss term to a point-to-plane distance loss term. The root mean square error (RMSE) in three experimental regions decreased from 0.149 m, 0.185 m, and 0.217 m to 0.038 m, 0.035 m, and 0.078 m. For comparative analysis, classical Non-Rigid ICP and TrICP (Trimmed Iterative Closest Point) algorithms are implemented using the same preprocessing procedures. The results demonstrate that the proposed method achieves better performance than both the Non-Rigid ICP and TrICP approaches in terms of geometric consistency. These results demonstrate the advantages of the proposed method in terms of both accuracy and robustness when applied to complex scenarios.