Improved Single-Layer Coding of Volumetric Data
Nahid Sheikhi-Pour, Mika Pesonen, Sebastian Schwarz, Vinod Kumar Malamal Vadakital · 2019
The Moving Picture Experts Group (MPEG) recently has developed a standard for compressing volumetric content known as video-based point cloud compression (V-PCC). The V-PCC standard is a multi-layer approach in which the point cloud information is mapped onto multiple image layers. The second layer is considered for including the points that were not projected into the first layer. Even though this dual-layer scheme provides better coding efficiency and reconstruction quality compared to the single-layer approach, but it is considered to be very complex due to high frame rate, memory buffer, execution time, etc. Hence, the single-layer scheme is considered in practical scenarios. This work attempts to improve the coding efficiency and reconstruction quality of the single-layer scheme. The proposed method in this work estimates the most important patches (MIPs) of the second layer and includes them in the patch generation of the first layer. The proposed method provided on average 9.8% and 5.7% bitrate reductions in all intra configuration in different quality metrics when compared to the single-layer scheme. Moreover, the proposed method in random access configuration has reduced the bitrate on average by 8.6% and 4.3% in different quality metrics compared to the single-layer scheme.