Beyond Illumination: Fine-Grained Detail Preservation in Extreme Dark Image Restoration

2025-08-05

AI Summary

To address severe structural degradation and noise corruption in extreme low-light images—particularly hindering faithful recovery of fine details (e.g., text and edges)—this paper proposes a two-stage frequency-domain enhancement framework. In the first stage, a residual Fourier-guided module restores global illumination. In the second stage, a gradient-aware Mamba module progressively reconstructs textures and sharp boundaries within the frequency-domain residual space, leveraging non-downsampled block-wise state modeling and channel correlation mechanisms. This design innovatively mitigates prior error accumulation and state decay commonly encountered in sequential modeling. Extensive experiments demonstrate significant improvements in detail fidelity across multiple benchmark datasets and downstream tasks—including edge detection and text recognition—while maintaining lightweight computational overhead. The method is modular and plug-and-play, seamlessly integrating into existing Fourier-domain enhancement pipelines.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Recovering fine-grained details in extremely dark images remains challenging due to severe structural information loss and noise corruption. Existing enhancement methods often fail to preserve intricate details and sharp edges, limiting their effectiveness in downstream applications like text and edge detection. To address these deficiencies, we propose an efficient dual-stage approach centered on detail recovery for dark images. In the first stage, we introduce a Residual Fourier-Guided Module (RFGM) that effectively restores global illumination in the frequency domain. RFGM captures inter-stage and inter-channel dependencies through residual connections, providing robust priors for high-fidelity frequency processing while mitigating error accumulation risks from unreliable priors. The second stage employs complementary Mamba modules specifically designed for textural structure refinement: (1) Patch Mamba operates on channel-concatenated non-downsampled patches, meticulously modeling pixel-level correlations to enhance fine-grained details without resolution loss. (2) Grad Mamba explicitly focuses on high-gradient regions, alleviating state decay in state space models and prioritizing reconstruction of sharp edges and boundaries. Extensive experiments on multiple benchmark datasets and downstream applications demonstrate that our method significantly improves detail recovery performance while maintaining efficiency. Crucially, the proposed modules are lightweight and can be seamlessly integrated into existing Fourier-based frameworks with minimal computational overhead. Code is available at https://github.com/bywlzts/RFGM.

Problem

• Recovering fine-grained details in extremely dark images
• Preserving intricate details and sharp edges in dark images
• Improving detail recovery performance while maintaining efficiency

Innovation

• Dual-stage approach for dark image detail recovery
• Residual Fourier-Guided Module for global illumination
• Patch and Grad Mamba for texture refinement