Perceptually Weighted Rate Distortion Optimization for Video-Based Point Cloud Compression.

Area of Science:

• Computer Vision
• Multimedia Compression
• Virtual and Augmented Reality

Background:

• Dynamic point clouds are crucial for immersive applications but suffer from large data volumes, hindering processing and transmission.
• Existing compression methods often struggle to balance data reduction with perceptual quality preservation.

Purpose of the Study:

• To develop an effective compression scheme for dynamic point clouds that minimizes perceptual distortion at a given bitrate.
• To introduce a novel objective metric for evaluating the perceptual quality of 3D point clouds.

Main Methods:

• A general framework for perceptually optimized Video-based Point Cloud Compression (V-PCC) was proposed.
• A multi-scale Projection based Point Cloud quality Metric (PPCM) was developed, involving 3D-to-2D projection, structural distortion measurement, and fusion.
• A Perceptually Weighted Rate-Distortion Optimization (PWRDO) scheme with Lagrange multiplier adaptation was integrated into V-PCC.

Main Results:

• The PPCM demonstrated higher consistency with human subjective scores than existing metrics.
• The PWRDO-based V-PCC scheme achieved significant average bitrate reductions (e.g., 13.52%) compared to the reference model.
• The computational overhead for PWRDO was negligible for both V-PCC encoder and decoder.

Conclusions:

• The proposed PPCM and PWRDO schemes offer a superior solution for dynamic point cloud compression.
• This approach effectively enhances coding efficiency and perceptual quality in V-PCC applications.
• The developed methods provide a valuable tool for optimizing 3D data handling in VR/AR.