Comparison and application of point cloud thinning algorithms for laser scanning cut tobacco accumulation

To solve the problem of redundancy in the original point cloud data obtained by laser scanning, which affected the efficiency of subsequent data processing and analysis, the applicability of laser scanning point cloud data thinning algorithms was investigated using random thinning, voxel segmentation thinning, Poisson density thinning, and curvature thinning. The results showed that the boundary of randomly thinned point cloud data was fuzzy, and the volume relative error was unstable. The point cloud thinned by voxel segmentation had a uniform distribution and preserved the feature shape completely. Poisson density thinning weakened features with higher density but achieved obvious simplification. Thinning based on curvature was easy to cause the loss of part of the surface shape, so it was necessary to reduce the thinning rate to make the volume error smaller. On the whole, the voxel thinning method was more suitable for the point cloud thinning of tobacco deposits. When the voxel side length was 10 mm, the volume relative error was less than 1%, and the thinning rate can reach ∼50%.