MEGA-PCC: A Mamba-based Efficient Approach for Joint Geometry and Attribute Point Cloud Compression

2025-12-26

AI Summary

To address the challenges of end-to-end optimization in joint geometric and attribute compression of point clouds—specifically, reliance on post-hoc recoloring and hand-crafted bit-rate allocation—this paper proposes the first Mamba-based, fully learnable end-to-end framework. Methodologically, it employs a shared encoder to produce a unified latent representation and two sequential decoders for geometry and attribute reconstruction. We introduce the Mamba Entropy Model (MEM), which jointly captures spatial and channel-wise dependencies to enable data-driven, adaptive bit-rate allocation. Our key contributions lie in pioneering the integration of Mamba into point cloud compression, unifying sequence modeling, joint implicit representation learning, and adaptive entropy coding, all optimized end-to-end under a rate-distortion objective. Experiments demonstrate state-of-the-art performance across multiple benchmarks, achieving superior rate-distortion trade-offs and inference speed, while enabling a more streamlined compression pipeline and efficient deployment.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Data Mining & Knowledge Management: Data Compression
• Computer Vision: Diffusion Models for Vision

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Joint compression of point cloud geometry and attributes is essential for efficient 3D data representation. Existing methods often rely on post-hoc recoloring procedures and manually tuned bitrate allocation between geometry and attribute bitstreams in inference, which hinders end-to-end optimization and increases system complexity. To overcome these limitations, we propose MEGA-PCC, a fully end-to-end, learning-based framework featuring two specialized models for joint compression. The main compression model employs a shared encoder that encodes both geometry and attribute information into a unified latent representation, followed by dual decoders that sequentially reconstruct geometry and then attributes. Complementing this, the Mamba-based Entropy Model (MEM) enhances entropy coding by capturing spatial and channel-wise correlations to improve probability estimation. Both models are built on the Mamba architecture to effectively model long-range dependencies and rich contextual features. By eliminating the need for recoloring and heuristic bitrate tuning, MEGA-PCC enables data-driven bitrate allocation during training and simplifies the overall pipeline. Extensive experiments demonstrate that MEGA-PCC achieves superior rate-distortion performance and runtime efficiency compared to both traditional and learning-based baselines, offering a powerful solution for AI-driven point cloud compression.

Problem

• Joint compression of point cloud geometry and attributes
• Eliminates need for recoloring and heuristic bitrate tuning
• Models long-range dependencies for improved probability estimation

Innovation

• Shared encoder for unified geometry and attribute latent representation
• Mamba-based entropy model capturing spatial and channel correlations
• End-to-end learning framework eliminating recoloring and heuristic tuning