UHDRes: Ultra-High-Definition Image Restoration via Dual-Domain Decoupled Spectral Modulation

2025-11-07

AI Summary

To address the challenge of efficiently restoring ultra-high-definition (UHD) images degraded by complex factors—including blur, haze, rain, fog, and low illumination—this paper proposes a lightweight dual-domain decoupled spectral modulation framework. Our method explicitly enhances frequency-domain magnitude while implicitly optimizing spatial-domain phase, enabling principled separation of degradation modeling across domains. We introduce a multi-scale spatial-spectral fusion mechanism and a shared gated feed-forward network to facilitate synergistic cross-domain feature interaction. By integrating spectral transforms, efficient frequency-domain modulation, multi-scale contextual aggregation, and adaptive gating, our approach substantially reduces model complexity. Evaluated on five UHD benchmarks, it achieves state-of-the-art performance with only 400K parameters, significantly lowering inference latency and memory footprint. The framework thus delivers an exceptional trade-off between restoration fidelity and computational efficiency.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Multimodal Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

Ultra-high-definition (UHD) images often suffer from severe degradations such as blur, haze, rain, or low-light conditions, which pose significant challenges for image restoration due to their high resolution and computational demands. In this paper, we propose UHDRes, a novel lightweight dual-domain decoupled spectral modulation framework for UHD image restoration. It explicitly models the amplitude spectrum via lightweight spectrum-domain modulation, while restoring phase implicitly through spatial-domain refinement. We introduce the spatio-spectral fusion mechanism, which first employs a multi-scale context aggregator to extract local and global spatial features, and then performs spectral modulation in a decoupled manner. It explicitly enhances amplitude features in the frequency domain while implicitly restoring phase information through spatial refinement. Additionally, a shared gated feed-forward network is designed to efficiently promote feature interaction through shared-parameter convolutions and adaptive gating mechanisms. Extensive experimental comparisons on five public UHD benchmarks demonstrate that our UHDRes achieves the state-of-the-art restoration performance with only 400K parameters, while significantly reducing inference latency and memory usage. The codes and models are available at https://github.com/Zhao0100/UHDRes.

Problem

• Restoring degraded ultra-high-definition images from blur and adverse conditions
• Addressing computational challenges in high-resolution image restoration
• Achieving efficient restoration with lightweight architecture and low latency

Innovation

• Dual-domain decoupled spectral modulation framework
• Lightweight spectrum-domain amplitude modulation
• Spatial-domain implicit phase restoration