Evaluation of Pointerless Sparse Voxel Octrees Encoding Schemes Using Huffman Encoding for Dense Volume Datasets Storage

Abstract

The paper deals with the problematics of the volume datasets representation using pointerless sparse voxel octrees hierarchical data structure. While this data structure, and derived hierarchical data structures based on the use of directed acyclic graphs, are popular in representation of the geometry of the three-dimensional scenes, their efficiency in the representation of other properties of voxels, e.g. color, is lower, due to the lower success in the lossless compression of the data. One of the solutions is to decompose volume dataset that comprises multi-bit values of its voxels, into bit planes and to use classical approach of their encoding to the sparse voxel hierarchical data structures. The paper evaluates this approach along with the different encoding schemes of pointerless sparse voxel octrees in case of multi-bit volume datasets that were obtained by non-invasive medical imaging techniques, including Computed Tomography and Magnetic Resonance Imaging. Five encoding schemes were leveraged in the paper and using test results, their lossless compression capabilities were evaluated.