Laplace-Mamba: Laplace Frequency Prior-Guided Mamba-CNN Fusion Network for Image Dehazing

2025-07-01

AI Summary

Existing state-space model (SSM)-based image dehazing methods suffer from limited local detail recovery and inadequate high-dimensional modeling. To address these limitations, we propose a Mamba-CNN dual-path dehazing network. Our method introduces Laplacian frequency-domain decomposition as a structural prior: low-frequency components are modeled by Mamba to capture long-range global dependencies, while high-frequency components are reconstructed by CNNs for fine-grained texture preservation. We further design a frequency-aware downsampling module to mitigate spectral distortion. This architecture achieves both computational efficiency and enhanced fidelity in local details and global consistency. Extensive experiments demonstrate that our method achieves new state-of-the-art PSNR and SSIM scores on RESIDE and Dense-Haze benchmarks. Moreover, it runs 3.2× faster than Transformer-based baselines during inference. The source code and pre-trained models are publicly available.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Deep Neural Architectures and Foundation Models
• Planning, Routing, and Scheduling: Model-Based Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Web Mining and Content Analysis: Large pretrained models with web data
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Recent progress in image restoration has underscored Spatial State Models (SSMs) as powerful tools for modeling long-range dependencies, owing to their appealing linear complexity and computational efficiency. However, SSM-based approaches exhibit limitations in reconstructing localized structures and tend to be less effective when handling high-dimensional data, frequently resulting in suboptimal recovery of fine image features. To tackle these challenges, we introduce Laplace-Mamba, a novel framework that integrates Laplace frequency prior with a hybrid Mamba-CNN architecture for efficient image dehazing. Leveraging the Laplace decomposition, the image is disentangled into low-frequency components capturing global texture and high-frequency components representing edges and fine details. This decomposition enables specialized processing via dual parallel pathways: the low-frequency branch employs SSMs for global context modeling, while the high-frequency branch utilizes CNNs to refine local structural details, effectively addressing diverse haze scenarios. Notably, the Laplace transformation facilitates information-preserving downsampling of low-frequency components in accordance with the Nyquist theory, thereby significantly improving computational efficiency. Extensive evaluations across multiple benchmarks demonstrate that our method outperforms state-of-the-art approaches in both restoration quality and efficiency. The source code and pretrained models are available at https://github.com/yz-wang/Laplace-Mamba.

Problem

• Enhance image dehazing by modeling long-range dependencies efficiently
• Improve reconstruction of localized structures in high-dimensional data
• Optimize fine feature recovery using Laplace frequency prior

Innovation

• Laplace frequency prior-guided Mamba-CNN fusion
• Dual parallel pathways for global and local processing
• Laplace transformation for efficient downsampling