Subjective Quality Assessment of V-PCC-Compressed Dynamic Point Clouds Degraded by Packet Losses

Area of Science:

• Computer Vision
• Multimedia Compression
• Data Transmission

Background:

• Dynamic point clouds are increasingly used in immersive applications.
• Compression is essential for efficient transmission and storage.
• Information loss during transmission can degrade reconstructed point cloud quality.

Purpose of the Study:

• To empirically investigate the impact of information loss on compressed dynamic point cloud quality.
• To assess the subjective quality of reconstructed point clouds under varying compression and packet loss conditions.
• To correlate subjective quality scores with objective quality metrics.

Main Methods:

• Dynamic point clouds compressed using MPEG V-PCC codec at 5 levels.
• Simulated packet losses at 0.5%, 1%, and 2% applied to sub-bitstreams.
• Subjective quality assessed via Mean Opinion Scores (MOS) by human observers in two labs.
• Statistical analysis of MOS scores and correlation with objective quality measures (FSIM, MSE, SSIM, PCQM).

Main Results:

• A 0.5% packet loss rate reduced subjective quality by 1-1.5 MOS units.
• Image-based metrics (FSIM, MSE, SSIM) and PCQM showed high correlation with subjective scores.
• Degradation of occupancy and geometry sub-bitstreams had a greater negative impact than attribute sub-bitstream degradation.

Conclusions:

• Dynamic point cloud bitstreams require adequate protection against packet loss.
• Objective quality metrics like PCQM and adapted image metrics can effectively predict subjective quality.
• Prioritizing the protection of occupancy and geometry data is crucial for maintaining perceived quality.