U-Net-Like Spiking Neural Networks for Single Image Dehazing
Huibin Li, Haoran Liu, Mingzhe Liu, Yulong Xiao, Peng Li, Guibin Zan
2026-01-07
Summary
This paper focuses on improving how computers can remove haze from images, making them clearer. It introduces a new method called DehazeSNN that aims to be both effective and efficient.
What's the problem?
Removing haze from images is tough because traditional methods don't always work well, and newer 'deep learning' approaches, while better, have drawbacks. Specifically, methods using Convolutional Neural Networks (CNNs) struggle to understand relationships between distant parts of an image, and methods using Transformers require a lot of computing power.
What's the solution?
The researchers created DehazeSNN, a new system that combines the best parts of a common image processing design called U-Net with a different type of neural network called a Spiking Neural Network (SNN). This combination allows the system to analyze images at different scales and handle both nearby and faraway relationships efficiently. They also added a special component, the Orthogonal Leaky-Integrate-and-Fire Block (OLIFBlock), to help different parts of the image communicate better, leading to improved haze removal with less computational effort.
Why it matters?
This work is important because it offers a way to remove haze from images that is both high-quality and doesn't require massive amounts of computing power. This means it could be used in more places, like on phones or in real-time applications, and the code is available for others to use and build upon.
Abstract
Image dehazing is a critical challenge in computer vision, essential for enhancing image clarity in hazy conditions. Traditional methods often rely on atmospheric scattering models, while recent deep learning techniques, specifically Convolutional Neural Networks (CNNs) and Transformers, have improved performance by effectively analyzing image features. However, CNNs struggle with long-range dependencies, and Transformers demand significant computational resources. To address these limitations, we propose DehazeSNN, an innovative architecture that integrates a U-Net-like design with Spiking Neural Networks (SNNs). DehazeSNN captures multi-scale image features while efficiently managing local and long-range dependencies. The introduction of the Orthogonal Leaky-Integrate-and-Fire Block (OLIFBlock) enhances cross-channel communication, resulting in superior dehazing performance with reduced computational burden. Our extensive experiments show that DehazeSNN is highly competitive to state-of-the-art methods on benchmark datasets, delivering high-quality haze-free images with a smaller model size and less multiply-accumulate operations. The proposed dehazing method is publicly available at https://github.com/HaoranLiu507/DehazeSNN.