When Color-Space Decoupling Meets Diffusion for Adverse-Weather Image Restoration

2025-09-21

AI Summary

Adverse weather image restoration (AWIR) suffers from unpredictable degradation patterns and poor generalization across diverse weather conditions. To address these challenges, this paper proposes LCDiff—a novel diffusion-based framework that uniquely integrates YCbCr color space decoupling with generative modeling. Specifically, it introduces a Lumina-Chroma Decomposition Network (LCDN) to disentangle luminance (Lumina) and chrominance (Chroma) components; a Luminance-Guided Diffusion Model (LGDM) then leverages the restored luminance as conditional input to reconstruct chrominance, enabling degradation-agnostic restoration. Furthermore, a dynamic timestep loss is incorporated to stabilize training without explicit degradation priors. Extensive experiments demonstrate significant improvements over state-of-the-art methods across multiple AWIR tasks—including dehazing, deraining, and desnowing—on standard benchmarks. Additionally, we release DriveWeather, a large-scale, multi-condition driving dataset covering diverse weather scenarios, to advance benchmarking and research in AWIR.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Large Multimodal Models (LMMs)
• Planning, Routing, and Scheduling: Planning with Language Models

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Large pretrained models with web data
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Adverse Weather Image Restoration (AWIR) is a highly challenging task due to the unpredictable and dynamic nature of weather-related degradations. Traditional task-specific methods often fail to generalize to unseen or complex degradation types, while recent prompt-learning approaches depend heavily on the degradation estimation capabilities of vision-language models, resulting in inconsistent restorations. In this paper, we propose extbf{LCDiff}, a novel framework comprising two key components: extit{Lumina-Chroma Decomposition Network} (LCDN) and extit{Lumina-Guided Diffusion Model} (LGDM). LCDN processes degraded images in the YCbCr color space, separately handling degradation-related luminance and degradation-invariant chrominance components. This decomposition effectively mitigates weather-induced degradation while preserving color fidelity. To further enhance restoration quality, LGDM leverages degradation-related luminance information as a guiding condition, eliminating the need for explicit degradation prompts. Additionally, LGDM incorporates a extit{Dynamic Time Step Loss} to optimize the denoising network, ensuring a balanced recovery of both low- and high-frequency features in the image. Finally, we present DriveWeather, a comprehensive all-weather driving dataset designed to enable robust evaluation. Extensive experiments demonstrate that our approach surpasses state-of-the-art methods, setting a new benchmark in AWIR. The dataset and code are available at: https://github.com/fiwy0527/LCDiff.

Problem

• Restoring images degraded by unpredictable adverse weather conditions
• Overcoming limitations of task-specific methods and prompt-learning approaches
• Enhancing restoration quality without explicit degradation prompts

Innovation

• YCbCr color space decomposition for degradation separation
• Luminance-guided diffusion model without explicit degradation prompts
• Dynamic time step loss for balanced feature recovery