DiffCom: Decoupled Sparse Priors Guided Diffusion Compression for Point Clouds

DiffCom presents a diffusion-based point cloud compression framework that decouples storage-efficient sparse priors from reconstruction latents, reducing latent redundancy while preserving geometric fidelity. By leveraging sparse anchors, inter-point local distributions, a conditional denoiser, and context-aware entropy coding, it achieves high-quality reconstruction and strong rate-distortion performance on ShapeNet and standard MPEG benchmarks, especially at low bitrates.

Recommended citation: Xiaoge Zhang, Zijie Wu, Mingtao Feng, Mehwish Nasim, Saeed Anwar, & Ajmal Mian. (2026). DiffCom: Decoupled sparse priors guided diffusion compression for point clouds. IEEE Transactions on Circuits and Systems for Video Technology (TCSVT) .
Download Paper