Lossless Compression of Point-Based 3 D Models

With the increasing sizes and complexity of the 3D models, it is an emerging demand to develop graphics compression techniques in order to efficiently store, transmit, and visualize such data in networked environments. Although there has been a very rich literature in graphics compression, most techniques developed so far have focused on compressing polygonal meshes, where even the state-of-the-art geometry coders for compressing the vertex information [3,6,7,10] require the use of the connectivity information and thus are not applicable to point-set compression. Among the relatively few results on pointset compression, the technique of [11] mainly focuses on multiresolution (rather than single-resolution) compression, and those of [4, 8] require resampling which alters the datasets and may not be allowed in some applications. Clearly, point-set compression is still in its early stage. In this paper, we present a single-resolution technique for lossless compression of point-based 3D models. An initial quantization step is not needed and we can achieve a truly lossless compression. The scheme can compress geometry information as well as attributes associated with the points. We employ a three-stage pipeline that uses several ideas including k-d-tree-like partitioning, minimum-spanning-tree modeling, and a two-layer modified Huffman coding technique based on an optimal alphabet partitioning approach using a greedy heuristic. We show that the proposed technique achieves excellent lossless compression results.