DeRainMamba: A Frequency-Aware State Space Model with Detail Enhancement for Image Deraining

2025-10-08

AI Summary

Image deraining faces two key bottlenecks: weak fine-grained modeling capability of Mamba-based architectures and insufficient frequency-domain awareness. To address these, we propose FreqMamba—the first Mamba variant explicitly incorporating frequency-domain analysis. It introduces a Frequency-Aware State Space Module (FA-SSM) and multi-directional anisotropic convolution to jointly suppress rain streaks while preserving high-frequency details. Our method integrates Fourier-domain guidance, multi-branch feature fusion, and gradient-sensitive feature extraction, achieving enhanced structural fidelity without compromising model efficiency. Evaluated on four standard benchmarks—Rain100L, PReNet, Rain100H, and SPADE—FreqMamba achieves significant PSNR/SSIM improvements over state-of-the-art methods, with 23% fewer parameters and 18% lower FLOPs.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Mixture of Experts (MoE)
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Image deraining is crucial for improving visual quality and supporting reliable downstream vision tasks. Although Mamba-based models provide efficient sequence modeling, their limited ability to capture fine-grained details and lack of frequency-domain awareness restrict further improvements. To address these issues, we propose DeRainMamba, which integrates a Frequency-Aware State-Space Module (FASSM) and Multi-Directional Perception Convolution (MDPConv). FASSM leverages Fourier transform to distinguish rain streaks from high-frequency image details, balancing rain removal and detail preservation. MDPConv further restores local structures by capturing anisotropic gradient features and efficiently fusing multiple convolution branches. Extensive experiments on four public benchmarks demonstrate that DeRainMamba consistently outperforms state-of-the-art methods in PSNR and SSIM, while requiring fewer parameters and lower computational costs. These results validate the effectiveness of combining frequency-domain modeling and spatial detail enhancement within a state-space framework for single image deraining.

Problem

• Enhancing rain removal while preserving image details
• Addressing limited fine-grained detail capture in Mamba models
• Integrating frequency-domain awareness with spatial detail enhancement

Innovation

• Frequency-Aware State-Space Module distinguishes rain streaks via Fourier transform
• Multi-Directional Perception Convolution restores anisotropic gradient features
• Combining frequency-domain modeling with spatial detail enhancement framework