Point cloud coding: techniques, applications and standardization

Abstract: 
3D Media is one of the key applications in the next generation of consumer electronics, networks and entertainment. In many exciting new applications like 3D teleportation, mixed real and virtual reality, self driving cars and many other use cases, compression and transport of 3D geometric data is rapidly becoming a bottleneck. In this tutorial we present advances and techniques for compression of geometric data like meshes or point clouds. We will reveal the requirements for designing a point cloud codec for immersive video and present the results of hybrid point cloud architecture developed for a large scale mixed reality system. Next we overview the standardization activities in MPEG on Point Cloud compression that attracted significant interest from the industry and scientific community. We detail the use case and tools involved for objective/subjective evaluation and the procedure and timeline of this standardization activity. Further we detail some interesting research directions based on combined video/point cloud architectures and techniques based on the emerging field of graph based signal processing.

Presenter
Rufael Mekuria was born in Amsterdam. He received the MSc in Electrical Engineering from TU Delft in 2011 and the PhD from VU University in 2017. He was a researcher at CWI in Amsterdam the Netherlands between 2011 and 2016. He has led the standardization activities related to tele-immersive coding in MPEG, and chairs the AhG on point cloud coding. He received the best paper award in ACM MMSys, and is a regular invited speaker, and has over 25 peer reviewed publications in ACM IEEE and SPIE venues and 30+ standardization contribution, mostly in the area of tele-immersive VR using geometry compression based on meshes and point clouds. He is co-founder of Point cloud compression B.V. a company developing point cloud compression technologies for the industry and related standards.