UniLDiff: Unlocking the Power of Diffusion Priors for All-in-One Image Restoration

  • 2025-08-04 07:22:07
  • Zihan Cheng, Liangtai Zhou, Dian Chen, Ni Tang, Xiaotong Luo, Yanyun Qu
  • 0

Abstract

All-in-One Image Restoration (AiOIR) has emerged as a promising yetchallenging research direction. To address the core challenges of diversedegradation modeling and detail preservation, we propose UniLDiff, a unifiedframework enhanced with degradation- and detail-aware mechanisms, unlocking thepower of diffusion priors for robust image restoration. Specifically, weintroduce a Degradation-Aware Feature Fusion (DAFF) to dynamically injectlow-quality features into each denoising step via decoupled fusion and adaptivemodulation, enabling implicit modeling of diverse and compound degradations.Furthermore, we design a Detail-Aware Expert Module (DAEM) in the decoder toenhance texture and fine-structure recovery through expert routing. Extensiveexperiments across multi-task and mixed degradation settings demonstrate thatour method consistently achieves state-of-the-art performance, highlighting thepractical potential of diffusion priors for unified image restoration. Our codewill be released.

 

Quick Read (beta)

loading the full paper ...