Performance assessment of point cloud compression

Abstract

Compression of dynamic point clouds for streaming and broadcast is a big challenge arising in mixed and virtual reality applications. To advance this area, common performance benchmarking and assessment is key as to compare different technologies consistently. This paper presents a methodology for performance assessment that was established in the MPEG point cloud compression activity in a collaboration of leading companies and academic institutes. It includes selected datasets, objective metrics and target coding rates corresponding to distinguishable subjective qualities and different use cases. The performance assessment methodology is applied to a compression framework on a large dataset developing an anchor performance benchmark. The benchmark results and methodology are used in the recent MPEG Call for proposals on point cloud coding to consistently compare and evaluate technologies.