Spatial and Frequency Domain Adaptive Fusion Network for Image Deblurring

2025-02-20

AI Summary

Existing image deblurring methods typically operate in either the spatial or frequency domain alone, struggling to simultaneously preserve local details and global structural coherence. To address this limitation, we propose the Spatial-Frequency Dual-domain Adaptive Fusion Network (SFDNet), which introduces— for the first time—a learnable low-pass filter-driven dynamic frequency-domain decomposition mechanism. Coupled with a Gated Cross-Attention Fusion Module (GFM), SFDNet enables adaptive, weighted collaboration between spatial and frequency features. Built upon NAFBlock as the backbone, the network integrates a Dynamic Frequency-Domain Generation Module (FDGM) and Cross-Attention Mechanism (CAM), significantly improving blur boundary reconstruction fidelity and texture consistency. Extensive experiments on GoPro and RealBlur benchmarks demonstrate state-of-the-art performance in PSNR and SSIM, validating both the effectiveness and generalizability of the dual-domain fusion paradigm.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

Image deblurring aims to reconstruct a latent sharp image from its corresponding blurred one. Although existing methods have achieved good performance, most of them operate exclusively in either the spatial domain or the frequency domain, rarely exploring solutions that fuse both domains. In this paper, we propose a spatial-frequency domain adaptive fusion network (SFAFNet) to address this limitation. Specifically, we design a gated spatial-frequency domain feature fusion block (GSFFBlock), which consists of three key components: a spatial domain information module, a frequency domain information dynamic generation module (FDGM), and a gated fusion module (GFM). The spatial domain information module employs the NAFBlock to integrate local information. Meanwhile, in the FDGM, we design a learnable low-pass filter that dynamically decomposes features into separate frequency subbands, capturing the image-wide receptive field and enabling the adaptive exploration of global contextual information. Additionally, to facilitate information flow and the learning of complementary representations. In the GFM, we present a gating mechanism (GATE) to re-weight spatial and frequency domain features, which are then fused through the cross-attention mechanism (CAM). Experimental results demonstrate that our SFAFNet performs favorably compared to state-of-the-art approaches on commonly used benchmarks.

Problem

• Fusing spatial and frequency domains
• Dynamic feature decomposition
• Adaptive image deblurring

Innovation

• Fuses spatial and frequency domains
• Employs gated feature fusion block
• Uses dynamic low-pass filtering