DehazeNet: An End-to-End System for Single Image Haze Removal

By Bolun Cai, Xiangmin Xu, K. Jia, Chunmei Qing, Dacheng Tao. IEEE Transactions on Image Processing, 2016

3,049 citations recorded by the source corpus.

Single image haze removal is a challenging ill-posed problem. Existing methods use various constraints/priors to get plausible dehazing solutions. The key to achieve haze removal is to estimate a medium transmission map for an input hazy image. In this paper, we propose a trainable end-to-end system called DehazeNet, for medium transmission estimation. DehazeNet takes a hazy image as input, and outputs its medium transmission map that is subsequently used to recover a haze-free image via atmospheric scattering model. DehazeNet adopts convolutional neural network-based deep architecture, whose layers are specially designed to embody the established assumptions/priors in image dehazing. Specifically, the layers of Maxout units are used for feature extraction, which can generate almost all haze-relevant features. We also propose a novel nonlinear activation function in DehazeNet, called bilateral rectified linear unit, which is able to improve the quality of recovered haze-free image. We establish connections between the components of the proposed DehazeNet and those used in existing methods. Experiments on benchmark images show that DehazeNet achieves superior performance over existing methods, yet keeps efficient and easy to use.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/a127899f71b6ebc03521d2f9538335ebf1f29d78

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

Benchmarking Single-Image Dehazing and Beyond (2017, methodology): “DCP [9] FVR [10] BCCR [11] GRM [12] CAP [13] NLD [14] DehazeNet [15] MSCNN [16] AOD-Net [17] PSNR 16.”
AOD-Net: All-in-One Dehazing Network (2017, methodology): “[3] proposed a trainable end-to-end model for medium transmission estimation, called DehazeNet.”
NH-HAZE: An Image Dehazing Benchmark with Non-Homogeneous Hazy and Haze-Free Images (2020, methodology): “[14] introduced DehazeNet, one of the first deep learned methods for image dehazing.”
Real-World Underwater Enhancement: Challenges, Benchmarks, and Solutions Under Natural Light (2019, background): “In the context of image dehazing, the convolutional neural networks (CNNs) may work in an end-to-end manner, which learns the direct mapping from a degraded image to the corresponding clear scene [40], [41], [42].”
Dual Residual Networks Leveraging the Potential of Paired Operations for Image Restoration (2019, neutral): “[5] 21.”
Dense-Haze: A Benchmark for Image Dehazing with Dense-Haze and Haze-Free Images (2019, methodology): “Not surprisingly, although they do not introduce additional distortions, the learning-based approaches of Ren et al [21] and Cai et al [20], trained using synthetic hazy dataset, are not able to remove the hazy (white) appearance.”
Restoring Vision in Adverse Weather Conditions With Patch-Based Denoising Diffusion Models (2022, background): “There has been an independent line of work focusing solely on image dehazing [1], [49], [50], where also similar GAN based generative solutions were”
PSD: Principled Synthetic-to-Real Dehazing Guided by Physical Priors (2021, methodology): “[3] proposed an end-to-end network to generate transmission estimates.”
Kindling the Darkness: A Practical Low-light Image Enhancer (2019, background): “With the emergence of deep learning, a number of low-level vision tasks have been benefited from deep models, such as [14], [15] for denoising, [16] for super-resolution, [17] for compression artifact removal and [18] for dehazing.”
DoveNet: Deep Image Harmonization via Domain Verification (2019, background): “Image-to-Image Translation: A variety of tasks that map an input image to a corresponding output image are collectively named image-to-image translation, such as image super-resolution [15, 16, 22], inpainting [33, 50], colorization [51, 21], denoising [26], de-blurring [46], dehazing [38, 3], demo-”
DEA-Net: Single Image Dehazing Based on Detail-Enhanced Convolution and Content-Guided Attention (2023, methodology): “Recently, with the rapid development of deep learning, Convolution Neural Network (CNN) based dehazing methods achieve superior performance [2], [3], [4], [5], [6].”
All-In-One Image Restoration for Unknown Corruption (2022, background): “To be exact, it does not rely a mathematical model that explicitly defines the relationship between the corrupted and clean images as existing methods [1, 37].”
A deep CNN method for underwater image enhancement (2017, methodology): “Thus, many previous work use local smoothness assumption to address color correction [13] [14] [15] [16] [17] [18] and haze removal [19] [20] [21] [22] [23] [24] problems, which assumes that the color distortion and”
Seeing Through Fog Without Seeing Fog: Deep Multimodal Sensor Fusion in Unseen Adverse Weather (2019, background): “Especially fog and haze removal from conventional intensity image data have been explored extensively [68, 71, 34, 54, 37, 7, 38, 47].”
MSR-net: Low-light Image Enhancement Using Deep Convolutional Network (2017, methodology): “[4] propose a trainable end-to-end system named DehazeNet, which takes a hazy image as input and outputs its medium transmission map that is subsequently used to recover a haze-free image via atmospheric scattering model.”
FFA-Net: Feature Fusion Attention Network for Single Image Dehazing (2019, methodology): “In view of the prevailing success of deep learning in image processing tasks and the availability of large image datasets, (Cai et al. 2016)”
GridDehazeNet: Attention-Based Multi-Scale Network for Image Dehazing (2019, methodology): “(2011), DehazeNet Cai et al. (2016), MSCNN Ren et al. (2016), AOD-Net Li et al. (2017) and GFN Ren et al. (2018). The DCP is a prior-based method and is regarded as the baseline in single image dehazing.”