RDEIC: Accelerating Diffusion-Based Extreme Image Compression with Relay Residual Diffusion

TL;DR

RDEIC introduces a novel diffusion-based image compression method that uses compressed feature initialization and residual diffusion to improve fidelity and efficiency at extremely low bitrates.

Contribution

It proposes a new approach combining compressed feature initialization and residual diffusion, along with a fixed-step fine-tuning strategy, to enhance image reconstruction quality and efficiency.

Findings

Achieves state-of-the-art visual quality in extreme image compression.

Outperforms existing diffusion-based methods in fidelity and efficiency.

Demonstrates effectiveness through extensive experiments.

Abstract

Diffusion-based extreme image compression methods have achieved impressive performance at extremely low bitrates. However, constrained by the iterative denoising process that starts from pure noise, these methods are limited in both fidelity and efficiency. To address these two issues, we present Relay Residual Diffusion Extreme Image Compression (RDEIC), which leverages compressed feature initialization and residual diffusion. Specifically, we first use the compressed latent features of the image with added noise, instead of pure noise, as the starting point to eliminate the unnecessary initial stages of the denoising process. Second, we directly derive a novel residual diffusion equation from Stable Diffusion's original diffusion equation that reconstructs the raw image by iteratively removing the added noise and the residual between the compressed and target latent features. In this way, we effectively combine the efficiency of residual diffusion with the powerful generative capability of Stable Diffusion. Third, we propose a fixed-step fine-tuning strategy to eliminate the discrepancy between the training and inference phases, thereby further improving the reconstruction quality. Extensive experiments demonstrate that the proposed RDEIC achieves state-of-the-art visual quality and outperforms existing diffusion-based extreme image compression methods in both fidelity and efficiency. The source code and pre-trained models are available at https://github.com/huai-chang/RDEIC.