A fast point cloud precision registration method based on spatial grid division

Point cloud precision registration is the most time-consuming link in registration, and the main factor that hinders point cloud registration from meeting the real-time requirement. In this paper we propose a fast point cloud precision registration method based on spatial grid division. By virtue of the spatial grid, the scattered point cloud is transformed into the spatial grid sequence. During the applicaton of the Iterative Closest Point (ICP) algorithm, we can get the neighborhood information of any point with the aid of the neighborhood information of the grid sequence to efficiently determine the nearest neighbor points. When calculating the nearest neighbor points, the grids don’t interfere with each other. Hence, we can use GPU parallel processing to accelerate the search of the nearest neighbor points. Spatial grid division and parallel processing greatly improve registration efficiency.