Abstract:Suspended particles in hazy medium absorb and scatter light, severely degrading imaging quality. Numerous single-image dehazing methods have been proposed to reconstruct clear images from hazy ones. However, most of them focus on increasing depth and width to improve dehazing performance, which incurs high computation and energy costs. To address this issue, we propose a lightweight spiking convolutional neural network (CNN) referred to as retina-inspired spiking CNN (RI-SCNN) for the reconstruction of hazy images. Unlike conventional dehazing techniques, first, our proposed network simulates the hierarchical structure and cellular function of the retina and devises five network modules to efficiently encode and extract image features through ON and OFF roads. Furthermore, the linear reconstruction mechanism is introduced to integrate the outputs from different roads, adaptively preserving regions with optimal details and constructing a comprehensive visual representation. Finally, by the transformed atmospheric scattering formula, our network can generate the dehazy image. Incorporating the microscale spiking mechanism of the brain, the entire network leverages discrete binary spike trains for information encoding and transmission, directly trained by spiking surrogate gradient learning on integrate-and-fire (IF) neurons. Experimental results demonstrate the superiority of the proposed RI-SCNN in terms of quantitative dehazing performance, qualitative visual effect, energy efficiency, and run speed. Considering its lightweight architecture with ultralow computation and energy costs, the network is encouraged to be deployed in the visual sensor hardware to improve overall performance.

Single UHD Image Dehazing Via Interpretable Pyramid Network

Rethinking Performance Gains in Image Dehazing Networks

Vision Transformers for Single Image Dehazing

Ultra-High-Definition Image Dehazing via Multi-Guided Bilateral Learning

Single Image Dehazing Algorithm Based on Pyramid Mutil-Scale Transposed Convolutional Network

U-shaped Vision Mamba for Single Image Dehazing

Densely Connected Pyramid Dehazing Network

DU-Net: A new double U-shaped network for single image dehazing

Efficient Dehazing with Recursive Gated Convolution in U-Net: A Novel Approach for Image Dehazing

SCPA-Net: Self-calibrated Pyramid Aggregation for Image Dehazing.

Lightweight single image dehazing network with residual feature attention

"Pyramid Deep dehazing": An unsupervised single image dehazing method using deep image prior

Pyramid Channel-based Feature Attention Network for Image Dehazing.

Single Image Dehazing Via Dual-Path Recurrent Network

Progressive Feature Fusion Network for Realistic Image Dehazing

USID-Net: Unsupervised Single Image Dehazing Network via Disentangled Representations

TransDehaze: transformer-enhanced texture attention for end-to-end single image dehaze

Parallel Cross Strip Attention Network for Single Image Dehazing

DehazeDCT: Towards Effective Non-Homogeneous Dehazing via Deformable Convolutional Transformer

Retina-Inspired Lightweight Spiking Convolutional Neural Network for Single-Image Dehazing

HDUD-Net: Heterogeneous Decoupling Unsupervised Dehaze Network