Efficient RAW Image Deblurring with Adaptive Frequency Modulation

2025-05-30

AI Summary

To address the challenges of frequency-dependent blur modeling and high computational cost in RAW image deblurring, this paper proposes the first frequency-domain neural network specifically designed for the RAW domain. Methodologically: (1) an adaptive spectral position modulation module is introduced to dynamically calibrate blur responses across frequency bands; (2) frequency-domain skip connections are incorporated to enhance high-frequency detail reconstruction; and (3) end-to-end frequency-domain optimization is performed directly in the RAW domain. On RAW deblurring benchmarks, our method surpasses state-of-the-art approaches, achieving significant improvements in PSNR and SSIM while reducing MACs by 37%. Moreover, when transferred to the sRGB domain, it maintains comparable or superior performance, demonstrating the generalizability and efficiency of frequency-domain modeling. This work establishes a principled framework for frequency-aware learning in low-level vision tasks operating on sensor-native data.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding

Application Category

• Responsible Web: Algorithmic accountability and transparency on the web
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

Image deblurring plays a crucial role in enhancing visual clarity across various applications. Although most deep learning approaches primarily focus on sRGB images, which inherently lose critical information during the image signal processing pipeline, RAW images, being unprocessed and linear, possess superior restoration potential but remain underexplored. Deblurring RAW images presents unique challenges, particularly in handling frequency-dependent blur while maintaining computational efficiency. To address these issues, we propose Frequency Enhanced Network (FrENet), a framework specifically designed for RAW-to-RAW deblurring that operates directly in the frequency domain. We introduce a novel Adaptive Frequency Positional Modulation module, which dynamically adjusts frequency components according to their spectral positions, thereby enabling precise control over the deblurring process. Additionally, frequency domain skip connections are adopted to further preserve high-frequency details. Experimental results demonstrate that FrENet surpasses state-of-the-art deblurring methods in RAW image deblurring, achieving significantly better restoration quality while maintaining high efficiency in terms of reduced MACs. Furthermore, FrENet's adaptability enables it to be extended to sRGB images, where it delivers comparable or superior performance compared to methods specifically designed for sRGB data. The code will be available at https://github.com/WenlongJiao/FrENet .

Problem

• Deblurring RAW images with frequency modulation
• Handling frequency-dependent blur efficiently
• Restoring high-frequency details in RAW images

Innovation

• RAW-to-RAW deblurring in frequency domain
• Adaptive Frequency Positional Modulation module
• Frequency domain skip connections for details