54 / 2023-03-30 22:37:02

Adaptive fusion method of point cloud from different platforms based on supervoxel

Laser point cloud,supervoxel,point cloud density,adaptive,point cloud fusion

Point cloud fusion between different platforms is a crucial aspect for enhancing the completeness of spatial information acquisition in large scenes, particularly in the context of urban 3D model construction. Aiming at the problems of redundancy, noise, and accuracy degradation caused by direct registration of point cloud data from different platforms, an adaptive point cloud data fusion method with supervoxel is proposed. Firstly, the high-precision point cloud data are selected as the reference point cloud (RPC), and a coarse-to-fine registration approach using RANSAC-ICP is applied to achieve the spatial registration of RPC and target point cloud (TPC). Next, the supervoxel is constructed for the registered RPC and TPC. Finally, within each corresponding supervoxel, redundant and noisy points are removed by considering the data quality and density distribution information of the RPC, thereby achieving adaptive fusion of point cloud data from different platforms. Experimental validation was carried out on the data acquired from three different platforms. The proposed method can improve the accuracy of point cloud fusion, reduce the total number of point clouds by about 30%, and achieve the adaptive fusion of point cloud data. Moreover, it effectively preserves the detailed features of the fused point cloud data, providing accurate data sources for urban 3D model construction.