Abstract:Single image dehazing is a highly challenging ill-posed problem. Existing methods including both prior-based and learning-based heavily rely on the conceptual simplified atmospheric scattering model by estimating the so-called medium transmission map and atmospheric light. However, the formation of haze in the real world is much more complicated and inaccurate estimations further degrade the dehazing performance with color distortion, artifacts and insufficient haze removal. Moreover, most dehazing networks treat spatial-wise and channel-wise features equally, but haze is practically unevenly distributed across an image, thus regions with different haze concentrations require different attentions. To solve these problems, we propose an end-to-end trainable densely connected residual spatial and channel attention network based on the conditional generative adversarial framework to directly restore a haze-free image from an input hazy image, without explicitly estimation of any atmospheric scattering parameters. Specifically, a novel residual attention module is proposed by combining spatial attention and channel attention mechanism, which could adaptively recalibrate spatial-wise and channel-wise feature weights by considering interdependencies among spatial and channel information. Such a mechanism allows the network to concentrate on more useful pixels and channels. Meanwhile, the dense network can maximize the information flow along features from different levels to encourage feature reuse and strengthen feature propagation. In addition, the network is trained with a multi-loss function, in which contrastive loss and registration loss are novel refined to restore sharper structures and ensure better visual quality. Experimental results demonstrate that the proposed method achieves the state-of-the-art performance on both public synthetic datasets and real-world images with more visually pleasing dehazed results.

Dual-branch feature fusion dehazing network via multispectral channel attention

From Coarse to Fine: A Two-Stage Network for Dense Haze removal

Adaptive feature fusion network based on boosted attention mechanism for single image dehazing

Hyperspectral and Multispectral Image Fusion Based on Deep Attention Network.

Dual-path dehazing network with spatial-frequency feature fusion

Accurate and lightweight dehazing via multi-receptive-field non-local network and novel contrastive regularization

Multi-Scale Attention Feature Enhancement Network for Single Image Dehazing

Feature Fusion Image Dehazing Network Based on Hybrid Parallel Attention

Multi-scale recurrent attention gated fusion network for single image dehazing

Adaptive Multi-Feature Attention Network for Image Dehazing

FFA-Net: Feature Fusion Attention Network for Single Image Dehazing

A Multi-Scale Dehazing Network with Dark Channel Priors

Multi-Stream Fusion Network With Generalized Smooth L1 Loss for Single Image Dehazing

Frequency and content dual stream network for image dehazing

TFFD-Net: an effective two-stage mixed feature fusion and detail recovery dehazing network

Multiscale Depth Fusion With Contextual Hybrid Enhancement Network for Image Dehazing

Multi-level Feature Interaction and Efficient Non-Local Information Enhanced Channel Attention for image dehazing

Multi-Scale Boosted Dehazing Network with Dense Feature Fusion

Residual Spatial and Channel Attention Networks for Single Image Dehazing

Self-Adaptation Feature Attention Network with Multi-Step Fusion for Single Image Dehazing

MCPNet: Multi-space color correction and features prior fusion for single-image dehazing in Non-homogeneous haze Scenarios