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

TL;DR

Laplace-Mamba introduces a hybrid framework combining Laplace frequency prior with Mamba-CNN architecture, effectively disentangling image components for superior dehazing performance and computational efficiency.

Contribution

The paper proposes a novel Laplace-Mamba framework that integrates Laplace frequency prior with dual-path processing for enhanced image dehazing.

Findings

Outperforms state-of-the-art methods in dehazing quality

Achieves higher computational efficiency through Laplace-based downsampling

Effectively handles diverse haze scenarios with dual-path architecture

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.