Fig 1 - uploaded by P. Duhamel
Content may be subject to copyright.
Overview of the MSVoxelDNN architecture with input block of size 64 and 3 scales. DS is the downsampling operation (max-pooling). The base resolution of size 8 is encoded using a VoxelDNN context model. The higher resolutions are predicted from lower resolution as well as encoded groups at the same scale. The predicted block probabilities on the right side are passed to an arithmetic coder for encoding voxel occupancies. The final bitstream is the concatenation of all bits at all scales.
Source publication
We propose a practical deep generative approach for lossless point cloud geometry compression, called MSVoxelDNN, and show that it significantly reduces the rate compared to the MPEG G-PCC codec. Our previous work based on autoregressive models (VoxelDNN) has a fast training phase, however, inference is slow as the occupancy probabilities are predi...
Contexts in source publication
Context 1
... procedure is applied recursively to lower resolutions until the lowest scale, which is encoded using VoxelDNN. Figure 1 shows the general scheme of our Multiscale VoxelDNN encoder (MSVoxelDNN). At each step of the pyramid, downsampling is obtained by applying a maxpooling operation of size 2 × 2 × 2 to the high resolution block, i.e., the resulting lower resolution voxel occupancy is one if at least one of the 8 higher resolution voxels is occupied. ...
Context 2
... procedure is applied recursively to lower resolutions until the lowest scale, which is encoded using VoxelDNN. Figure 1 shows the general scheme of our Multiscale VoxelDNN encoder (MSVoxelDNN). At each step of the pyramid, downsampling is obtained by applying a maxpooling operation of size 2 × 2 × 2 to the high resolution block, i.e., the resulting lower resolution voxel occupancy is one if at least one of the 8 higher resolution voxels is occupied. ...
Similar publications
Point cloud compression is an essential task for practical applications using point clouds. Most of the previous approaches rely on octree compression which involves voxelization in the coding itself. Distortions derived from voxelization can be reduced without increasing the bitrate by post-processing. In this paper, we propose a super-resolution...
Due to the often substantial size of the real-world point cloud data, efficient transmission and storage have become critical concerns. Point cloud compression plays a decisive role in addressing these challenges. Recognizing the importance of capturing global information within point cloud data for effective compression, many existing point cloud...
Point clouds are becoming essential in key applications with advances in capture technologies leading to large volumes of data. Compression is thus essential for storage and transmission. In this work, the state of the art for geometry and attribute compression methods with a focus on deep learning based approaches is reviewed. The challenges faced...