DUGAE: Unified Geometry and Attribute Enhancement via Spatiotemporal Correlations for G-PCC Compressed Dynamic Point Clouds

2026-03-27

AI Summary

Existing post-decoding enhancement methods for point clouds primarily target static data and struggle to effectively exploit the spatiotemporal correlations inherent in dynamic point clouds. This work proposes DUGAE, a unified framework for joint geometric and attribute enhancement of compressed dynamic point clouds—the first to integrate both aspects within a single architecture. DUGAE explicitly models inter-frame spatiotemporal redundancy through a feature-domain geometry/attribute motion compensation mechanism (GMC/AMC), a detail-aware k-nearest neighbor recoloring module (DA-KNN), and sparse convolutions. Experimental results demonstrate that DUGAE significantly outperforms G-PCC GeS-TM v10 across multiple standard datasets, achieving a 11.03 dB gain in geometry BD-PSNR (93.95% bitrate saving) and a 4.23 dB improvement in luminance BD-PSNR (66.61% bitrate saving), while also surpassing the performance of V-PCC.

Abstract

Existing post-decoding quality enhancement methods for point clouds are designed for static data and typically process each frame independently. As a result, they cannot effectively exploit the spatiotemporal correlations present in point cloud sequences.We propose a unified geometry and attribute enhancement framework (DUGAE) for G-PCC compressed dynamic point clouds that explicitly exploits inter-frame spatiotemporal correlations in both geometry and attributes. First, a dynamic geometry enhancement network (DGE-Net) based on sparse convolution (SPConv) and feature-domain geometry motion compensation (GMC) aligns and aggregates spatiotemporal information. Then, a detail-aware k-nearest neighbors (DA-KNN) recoloring module maps the original attributes onto the enhanced geometry at the encoder side, improving mapping completeness and preserving attribute details. Finally, a dynamic attribute enhancement network (DAE-Net) with dedicated temporal feature extraction and feature-domain attribute motion compensation (AMC) refines attributes by modeling complex spatiotemporal correlations. On seven dynamic point clouds from the 8iVFB v2, Owlii, and MVUB datasets, DUGAE significantly enhanced the performance of the latest G-PCC geometry-based solid content test model (GeS-TM v10). For geometry (D1), it achieved an average BD-PSNR gain of 11.03 dB and a 93.95% BD-bitrate reduction. For the luma component, it achieved a 4.23 dB BD-PSNR gain with a 66.61% BD-bitrate reduction. DUGAE also improved perceptual quality (as measured by PCQM) and outperformed V-PCC. Our source code will be released on GitHub at: https://github.com/yuanhui0325/DUGAE

Problem

• point cloud
• spatiotemporal correlation
• quality enhancement
• dynamic point cloud
• G-PCC

Innovation

• spatiotemporal correlation
• geometry and attribute enhancement
• motion compensation
• dynamic point clouds
• G-PCC