A Geometry-aware Framework for Compressing 3D Mesh Textures

Abstract

This paper proposes a novel prediction tool for improving the compression performance of texture atlases. This algorithm, called Geometry-Aware (GA) intra coding, takes advantage of the topology of the associated 3D meshes, in order to reduce the redundancies in the texture map. For texture processing, the concept of the conventional intra prediction, used in video compression, has been adapted to consider neighboring information on the 3D surface. We have also studied how this prediction tool can be integrated into a complete coding solution. In particular, a block scanning strategy and a graph-based transform for residual coding have been proposed. Results show that the knowledge of the mesh topology significantly improves the compression efficiency of texture atlases 1 .

A geometry-aware compression of 3D mesh texture with random access

“…Let us define the neighborhood of the block with index i as the set of adjacent block indices, N (i) = {j | a i ∼ a j }. In [10] as in all conventional coders, the set of blocks whose index are in N (i) and that are already coded are used to generate a prediction [11], denoted byâ i . A residue a i −â i is then computed and coded with an arithmetic coder.…”

“…Here we give more details on the different aforementioned steps. The proposed coder builds upon [10], and adds to it the random accessibility at the decoder side. For the sake of clarity, some useful information from [10] are recalled in the following.…”

“…In a classical 2D image, N (i) is simply composed of the 4-connected neighbor indices placed on the previous/next column/row. As discussed in our previous work [10], for some blocks, some textural information that are neighbors in the 3D model can be split into distant blocks in the atlas. Thanks to the mesh topology, we retrieve this proximity and describe it in N (i) in addition to its 4-connected neighbors.…”

“…A geometry-aware atlas coder has been proposed in [10] to circumvent the first limitation. This coder takes into account the geometry information at several key stages of the compression algorithm, i.e.…”

“…As a consequence, the block decoding order can be decided at the decoder side, once the request is known, such that only the visible blocks are transmitted. In other words, the proposed coder brings random access ability with a reasonable storage overhead compared to [10], but it significantly reduces the transmission rate. Even better, the transmission efficiency is as if the encoder already knew which part would be requested by the user, and can thus be considered as optimal.…”