The Comparison of Compression and Conversion Performance for Real-time Volumetric Video Streaming by Using Dynamic Point Cloud and Mesh Coding

In volumetric video content, dynamic 3D objects containing frame information, akin to a video, are utilized to depict the movement of 3D objects. Due to their larger data size compared to static 3D objects, efficient compression techniques are imperative for dynamic 3D objects. Consequently, the moving picture experts group (MPEG) has undertaken research on video-based dynamic mesh coding (V-DMC) for compressing dynamic mesh, and video-based point cloud compression (V-PCC) for compressing dynamic point cloud. Moreover, mesh and point cloud serve as data structures representing 3D object in virtual reality, each with distinct advantages and disadvantages. Hence, advancements in volumetric video can be anticipated through the conversion of dynamic point cloud obtained from real world into dynamic meshes, which offer rendering advantages. Enabling real-time streaming of volumetric video using dynamic mesh necessitates not only the conversion between dynamic point cloud and dynamic mesh but also efficient compression techniques tailored for dynamic point cloud or dynamic mesh. This paper conducts experiments involving conversion and compression from dynamic point cloud to dynamic mesh using a poisson surface reconstruction (PSR) based converter and international compression standards for dynamic mesh and dynamic point cloud. Furthermore, subjective performance evaluations are conducted on the experimental results, comparing the dynamic meshes generated using V-DMC and V-PCC.