Traffic Image Restoration under Adverse Weather via Frequency-Aware Mamba

2025-12-03

AI Summary

Traffic image restoration under adverse weather conditions suffers from insufficient modeling of frequency-domain priors. To address this, we propose Frequency-Aware Mamba (FAM), the first framework to integrate state-space models (specifically, Mamba) into frequency-domain image restoration. FAM introduces a bidirectional 2D frequency-adaptive scanning mechanism that dynamically prioritizes wavelet high-frequency regions; a dual-branch feature extraction and prior-guided module that jointly leverages complementary spatial- and frequency-domain representations; and wavelet-domain high-frequency residual learning to enhance texture and detail reconstruction. Extensive experiments on diverse adverse-weather datasets (e.g., fog and rain) demonstrate that FAM significantly outperforms leading spatial-domain methods, achieving state-of-the-art performance in both quantitative metrics (PSNR/SSIM) and visual quality. This validates the effectiveness and advancement of frequency-aware sequential modeling for traffic image restoration.

Technology Category

• Computer Vision: Multi-modal Vision
• Planning, Routing, and Scheduling: Model-Based Reasoning
• Application Domains: Transportation

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
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Traffic image restoration under adverse weather conditions remains a critical challenge for intelligent transportation systems. Existing methods primarily focus on spatial-domain modeling but neglect frequency-domain priors. Although the emerging Mamba architecture excels at long-range dependency modeling through patch-wise correlation analysis, its potential for frequency-domain feature extraction remains unexplored. To address this, we propose Frequency-Aware Mamba (FAMamba), a novel framework that integrates frequency guidance with sequence modeling for efficient image restoration. Our architecture consists of two key components: (1) a Dual-Branch Feature Extraction Block (DFEB) that enhances local-global interaction via bidirectional 2D frequency-adaptive scanning, dynamically adjusting traversal paths based on sub-band texture distributions; and (2) a Prior-Guided Block (PGB) that refines texture details through wavelet-based high-frequency residual learning, enabling high-quality image reconstruction with precise details. Meanwhile, we design a novel Adaptive Frequency Scanning Mechanism (AFSM) for the Mamba architecture, which enables the Mamba to achieve frequency-domain scanning across distinct subgraphs, thereby fully leveraging the texture distribution characteristics inherent in subgraph structures. Extensive experiments demonstrate the efficiency and effectiveness of FAMamba.

Problem

• Restores traffic images in adverse weather conditions
• Integrates frequency guidance with sequence modeling
• Enhances local-global interaction via adaptive frequency scanning

Innovation

• Integrates frequency guidance with sequence modeling
• Uses bidirectional 2D frequency-adaptive scanning for local-global interaction
• Employs wavelet-based high-frequency residual learning for detail refinement