Fast compression and access of LiDAR point clouds using wavelets

To compress, access, visualize and analyze large 3D point clouds, they are often converted to Digital Surface Models, either as raster grids or Triangulated Irregular Networks. This paper proposes an approach, which works directly on the points as they were recorded during data capture. There is usually a strong correlation between successively recorded points. This correlation is used to compress the point clouds using wavelet transforms. The characteristics of wavelet coefficients are used to access areas of progressively higher resolution and quality. The detail coefficients can also be used for 3D analysis and reconstruction.