Low-bitrate Volumetric Video Streaming with Depth Image

Junquan Zhong,Haodan Zhang,Quanlu Jia,Jiangkai Wu,Peiheng Wang,Haoyang Wang,Liming Liu,Xinggong Zhang,Zongming Guo

DOI: https://doi.org/10.1145/3672196.3673397

2024-01-01

Abstract:Volumetric video, which is six-degrees-of-freedom (6DoF), serves an immersive viewing experience. However, the high bandwidth required for streaming volumetric video over networks presents challenges for its flourishing. To address this issue, we propose a novel streaming system that leverages depth image to convert 3D volumetric videos into 2D videos with low bitrate and high fps. Our system addresses two technical challenges when using depth image. Specifically, we use nonlinear regression to solve the virtual camera placement problem and the problem of how to allocate bitrate between color image and depth map when encoding. Evaluations show that our system outperforms existing volumetric video streaming methods and saves 50%-99% bandwidth under the same level of quality.