A joint team from the Chinese Academy of Sciences, Tongji University and Ningbo University has developed a breakthrough point cloud compression technology, TSC-PCAC, which significantly improves the compression efficiency and processing speed of point cloud data and solves the massive problem faced by 3D applications such as AR/VR. Data processing challenges. This technology is based on end-to-end voxel Transformer and sparse convolution, using a two-stage compression architecture to effectively reduce data redundancy, and optimizes inter-channel correlation through the innovative TSCM channel context module to further improve compression efficiency. This technology has achieved significant breakthroughs in data compression rate and processing speed, providing strong technical support for the development of 3D applications.
In the context of the current rapid development of 3D vision technology, point cloud, as a key data form for virtual reality and augmented reality, faces huge transmission and storage challenges. A high-quality point cloud may contain millions of data points, each carrying multi-dimensional information such as location, color, and transparency. The processing efficiency of these massive data directly affects the popularity of 3D applications.
To address this problem, the research team developed a point cloud attribute compression technology (TSC-PCAC) based on end-to-end voxel Transformer and sparse convolution. The core of this technology lies in its unique two-stage compression architecture: the first stage focuses on the extraction and modeling of local features of point clouds, and the second stage captures global features through a larger receptive field, effectively reducing data redundancy. .
The research team also innovatively designed a channel context module based on TSCM, which significantly improved data compression efficiency by optimizing the correlation between channels. Experimental data shows that compared with existing mainstream technologies, TSC-PCAC has achieved significant improvements in data compression rate: 38.53% higher than Sparse-PCAC, 21.30% higher than NF-PCAC, and 21.30% higher than G-PCAC. PCC v23 improved by 11.19%. Even more impressive is that its processing speed has also achieved a qualitative leap, with encoding and decoding times reduced by 97.68% and 98.78% respectively.
This breakthrough achievement not only solves the key pain points in point cloud data processing, but also lays an important foundation for the further development of 3D applications such as AR/VR. The research team stated that it will continue to explore deep network technology with higher compression ratios in the future and work on a unified processing solution for geometry and attribute encoding.
Paper address: https://arxiv.org/html/2407.04284v1
The successful development of TSC-PCAC technology marks significant progress in point cloud compression technology and provides strong technical support for the popularization and development of 3D applications such as AR/VR. It is expected to be widely used in more fields in the future.