dorsal/arxiv
View SchemaTowards Efficient Low-rate Image Compression with Frequency-aware Diffusion Prior Refinement
| Authors | Yichong Xia, Yimin Zhou, Jinpeng Wang, Bin Chen |
|---|---|
| Categories | |
| ArXiv ID | 2601.10373vv1 |
| URL | https://arxiv.org/abs/2601.10373 |
| License | http://creativecommons.org/licenses/by/4.0/ |
Abstract
Recent advancements in diffusion-based generative priors have enabled visually plausible image compression at extremely low bit rates. However, existing approaches suffer from slow sampling processes and suboptimal bit allocation due to fragmented training paradigms. In this work, we propose Accelerate \textbf{Diff}usion-based Image Compression via \textbf{C}onsistency Prior \textbf{R}efinement (DiffCR), a novel compression framework for efficient and high-fidelity image reconstruction. At the heart of DiffCR is a Frequency-aware Skip Estimation (FaSE) module that refines the $\epsilon$-prediction prior from a pre-trained latent diffusion model and aligns it with compressed latents at different timesteps via Frequency Decoupling Attention (FDA). Furthermore, a lightweight consistency estimator enables fast \textbf{two-step decoding} by preserving the semantic trajectory of diffusion sampling. Without updating the backbone diffusion model, DiffCR achieves substantial bitrate savings (27.2\% BD-rate (LPIPS) and 65.1\% BD-rate (PSNR)) and over $10\times$ speed-up compared to SOTA diffusion-based compression baselines.
{
"annotation_id": "da7c1c53-e43f-4022-8765-bf3f0b098344",
"date_created": "2026-02-17T05:53:24.170000Z",
"date_modified": "2026-02-17T05:53:24.170000Z",
"file_hash": "505bdb67a403d4be43dd5614fb587394b66692018f5182369921eb58c7bc6661",
"private": false,
"record": {
"abstract": "Recent advancements in diffusion-based generative priors have enabled visually plausible image compression at extremely low bit rates. However, existing approaches suffer from slow sampling processes and suboptimal bit allocation due to fragmented training paradigms. In this work, we propose Accelerate \\textbf{Diff}usion-based Image Compression via \\textbf{C}onsistency Prior \\textbf{R}efinement (DiffCR), a novel compression framework for efficient and high-fidelity image reconstruction. At the heart of DiffCR is a Frequency-aware Skip Estimation (FaSE) module that refines the $\\epsilon$-prediction prior from a pre-trained latent diffusion model and aligns it with compressed latents at different timesteps via Frequency Decoupling Attention (FDA). Furthermore, a lightweight consistency estimator enables fast \\textbf{two-step decoding} by preserving the semantic trajectory of diffusion sampling. Without updating the backbone diffusion model, DiffCR achieves substantial bitrate savings (27.2\\% BD-rate (LPIPS) and 65.1\\% BD-rate (PSNR)) and over $10\\times$ speed-up compared to SOTA diffusion-based compression baselines.",
"arxiv_id": "2601.10373",
"authors": [
"Yichong Xia",
"Yimin Zhou",
"Jinpeng Wang",
"Bin Chen"
],
"categories": [
"cs.CV",
"cs.AI",
"cs.MM"
],
"license": "http://creativecommons.org/licenses/by/4.0/",
"title": "Towards Efficient Low-rate Image Compression with Frequency-aware Diffusion Prior Refinement",
"url": "https://arxiv.org/abs/2601.10373",
"version": "v1"
},
"schema_id": "dorsal/arxiv",
"source": {
"execution_id": "f7aa480a-d749-497b-8bce-eddb4db706bc",
"id": "arXiv Dataset",
"type": "Model",
"variant": "snapshot-2026-01-17",
"version": "0.1.0"
},
"user_id": 1000002
}