Unified Geometry and Color Compression Framework for Point Clouds via Generative Diffusion Priors

2025-03-23

AI Summary

To address the limitations of disjoint geometric and color modeling and poor generalization in colored point cloud compression, this paper proposes the first end-to-end geometric-color joint compression framework leveraging generative diffusion priors. Departing from conventional decoupled processing paradigms, our method exploits a pre-trained diffusion model: via prompt tuning, it generates sparse seed points, which are then iteratively refined through denoising sampling to reconstruct the complete colored point cloud. Crucially, no dataset-specific fine-tuning is required—enabling zero-shot cross-distribution generalization. Evaluated on object- and indoor-scene benchmarks, our approach achieves 2.1–3.8 dB PSNR gains over state-of-the-art methods at equivalent bitrates, jointly optimizing compression ratio and reconstruction fidelity. The core innovation lies in the first integration of diffusion priors into point cloud compression, establishing a unified implicit representation that jointly encodes geometry and color.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Deep Generative Models & Autoencoders
• Data Mining & Knowledge Management: Data Compression

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Security and Privacy: Data transparency and provenance

Abstract

With the growth of 3D applications and the rapid increase in sensor-collected 3D point cloud data, there is a rising demand for efficient compression algorithms. Most existing learning-based compression methods handle geometry and color attributes separately, treating them as distinct tasks, making these methods challenging to apply directly to point clouds with colors. Besides, the limited capacities of training datasets also limit their generalizability across points with different distributions. In this work, we introduce a test-time unified geometry and color compression framework of 3D point clouds. Instead of training a compression model based on specific datasets, we adapt a pre-trained generative diffusion model to compress original colored point clouds into sparse sets, termed 'seeds', using prompt tuning. Decompression is then achieved through multiple denoising steps with separate sampling processes. Experiments on objects and indoor scenes demonstrate that our method has superior performances compared to existing baselines for the compression of geometry and color.

Problem

• Efficient compression for 3D colored point clouds
• Unified handling of geometry and color attributes
• Generalizability across diverse point distributions

Innovation

• Unified compression for geometry and color attributes
• Uses pre-trained generative diffusion model
• Prompt tuning for seed generation and denoising