Denoising Diffusion Probabilistic Model for Point Cloud Compression at Low Bit-Rates

2025-05-19

AI Summary

To address lossy point cloud compression under bandwidth-constrained, ultra-low-bitrate conditions, this paper introduces denoising diffusion probabilistic models (DDPMs) into point cloud compression for the first time, coupled with a learnable vector quantization (VQ)-based conditional coding mechanism to significantly improve structural fidelity at extremely low bitrates. The proposed method integrates a PointNet-based encoder, VQ entropy modeling, and a DDPM-driven generative prior, trained end-to-end via rate-distortion optimization. Evaluated on ShapeNet and ModelNet40, it achieves up to 2.3 dB PSNR gain over MPEG G-PCC and state-of-the-art methods in the sub-0.1 bpp regime, demonstrating substantial improvements in both reconstruction quality and bitrate efficiency. The implementation is publicly available.

Technology Category

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

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• 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

Abstract

Efficient compression of low-bit-rate point clouds is critical for bandwidth-constrained applications. However, existing techniques mainly focus on high-fidelity reconstruction, requiring many bits for compression. This paper proposes a"Denoising Diffusion Probabilistic Model"(DDPM) architecture for point cloud compression (DDPM-PCC) at low bit-rates. A PointNet encoder produces the condition vector for the generation, which is then quantized via a learnable vector quantizer. This configuration allows to achieve a low bitrates while preserving quality. Experiments on ShapeNet and ModelNet40 show improved rate-distortion at low rates compared to standardized and state-of-the-art approaches. We publicly released the code at https://github.com/EIDOSLAB/DDPM-PCC.

Problem

• Efficient compression of low-bit-rate point clouds
• Preserving quality while achieving low bitrates
• Improving rate-distortion at low rates

Innovation

• Denoising Diffusion Probabilistic Model for compression
• PointNet encoder with learnable vector quantizer
• Improved rate-distortion at low bitrates