Binary voxel map compression based on quadtree and octree subdivision, using arbitrary and non-power-of-2 dimensions

摘要

This paper presents an approach to store binary voxel maps of arbitrary, non-power-of-2 dimensions, using 2D and 3D spatial subdivision. The two data structures used are quadtrees and octrees. The approach presented is applied on voxel maps of bone tissue that has been microCT scanned at a resolution of a few microns. Due to space requirements all source data may not be expanded to cover sizes that are a power of 2, which would greatly simplify the subdivision approach. The 2D subdivision data structure is applied on a per z-level, while the 3D subdivision is applied on the entire voxel map. Either of the two approaches compresses the initial data, although not at the same level. The 2D approach allows selective decompression of each z-layer, without processing the remaining voxel map.