Diffusion-OAMP for Joint Image Compression and Wireless Transmission

2026-04-30

AI Summary

This work addresses the critical challenge of jointly optimizing image compression and wireless transmission in practical communication systems, where existing methods often lack generalizable restoration capabilities. The authors formulate the problem as an equivalent linear system and propose Diffusion-OAMP, a novel framework that, for the first time, integrates a pre-trained diffusion model into the Orthogonal Approximate Message Passing (OAMP) algorithm without requiring additional training. Within this framework, the OAMP linear module produces pseudo-additive white Gaussian noise (AWGN) observations, while the diffusion model acts as a nonlinear estimator guided by a signal-to-noise ratio matching rule, effectively fusing multiple generative priors. Experimental results demonstrate that the proposed method consistently outperforms conventional approaches across various compression ratios and noise levels, confirming its efficacy and robustness.

Abstract

Joint image compression and wireless transmission remain relatively underexplored compared to generic image restoration, despite its importance in practical communication systems. We formulate this problem under an equivalent linear model, and propose Diffusion-OAMP, a training-free reconstruction framework that embeds a pre-trained diffusion model into the OAMP algorithm. In Diffusion-OAMP, the OAMP linear estimator produces pseudo-AWGN observations, while the diffusion model serves as a nonlinear estimator under an SNR-matching rule. This framework offers a way to incorporate multiple generative priors into OAMP. Experiments with varying compression ratios and noise levels show that Diffusion-OAMP performs favorably against classic methods in the evaluated settings.

Problem

• image compression
• wireless transmission
• joint optimization
• communication systems

Innovation

• Diffusion-OAMP
• joint image compression and transmission
• OAMP
• diffusion model
• SNR-matching