Real Hazy Images Research Articles

PSD-ELGAN: A pseudo self-distillation based CycleGAN with enhanced local adversarial interaction for single image dehazing

Compared to pixel-level content loss, domain-level style loss in CycleGAN-based dehazing algorithms just imposes relatively soft constraints on the intermediate translated images, resulting in struggling to accurately model haze-free features from real hazy scenes. Furthermore, globally perceptual discriminator may misclassify real hazy images with significant scene depth variations as clean style, thereby resulting in severe haze residue. To address these issues, we propose a pseudo self-distillation based CycleGAN with enhanced local adversarial interaction for image dehazing, termed as PSD-ELGAN. On the one hand, we leverage the characteristic of CycleGAN to generate pseudo image pairs during training. Knowledge distillation is employed in this unsupervised framework to transfer the informative high-quality features from the self-reconstruction network of real clean images to the dehazing generator of paired pseudo hazy images, which effectively improves its haze-free feature representation ability without increasing network parameters. On the other hand, in the output of dehazing generator, four non-uniform image patches severely affected by residual haze are adaptively selected as input samples. The local discriminator could easily distinguish their hazy style, thereby further compelling the dehazing generator to suppress haze residues in such regions, thus enhancing its dehazing performance. Extensive experiments show that our PSD-ELGAN can achieve promising results and better generality across various datasets.

Remote sensing image dehazing using generative adversarial network with texture and color space enhancement

Remote sensing is gradually playing an important role in the detection of ground information. However, the quality of remote-sensing images has always suffered from unexpected natural conditions, such as intense haze phenomenon. Recently, convolutional neural networks (CNNs) have been applied to deal with dehazing problems, and some important findings have been obtained. Unfortunately, the performance of these classical CNN-based methods still needs further enhancement owing to their limited feature extraction capability. As a critical branch of CNNs, the generative adversarial network (GAN), composed of a generator and discriminator, has become a hot research topic and is considered a feasible approach to solving the dehazing problems. In this study, a novel dehazed generative adversarial network (GAN) is proposed to reconstruct the clean images from the hazy ones. For the generator network of the proposed GAN, the color and luminance feature extraction module and the high-frequency feature extraction module aim to extract multi-scale features and color space characteristics, which help the network to acquire texture, color, and luminance information. Meanwhile, a color loss function based on hue saturation value (HSV) is also proposed to enhance the performance in color recovery. For the discriminator network, a parallel structure is designed to enhance the extraction of texture and background information. Synthetic and real hazy images are used to check the performance of the proposed method. The experimental results demonstrate that the performance can significantly improve the image quality with a significant increment in peak-signal-to-noise ratio (PSNR). Compared with other popular methods, the dehazing results of the proposed method closely resemble haze-free images.

Cascaded combination of total variation regularization and contrast limited adaptive histogram equalization based image dehazing

ABSTRACT Haze and fog can significantly reduce the visibility of distant objects by scattering light and creating blurry, washed-out images lacking in detail. Image dehazing is a technique used in computer vision to enhance the clarity of such obscured images. This study examines the effectiveness of various dehazing methods using real hazy images and synthetic images with artificially created haze. Performance metrics such as PSNR, SSIM, MSE, CII, and computation time are used to evaluate the proposed method. The evaluation is carried out on datasets like RESIDE, GMAN, O-Haze, I-Haze, NH-Haze, and Dense haze to compare the proposed method with existing models. The proposed method attains 27.46%, 20.63% and 21.09% higher PSNR and 12.36%, 23.95% and 36.12% lower Natural Image Quality Evaluator for RESIDE dataset when analyzed to the existing models, such as strategic method towards contrast enhancement by 2D histogram equalization under TV decomposition (CE-TDHE-TVD), multiple level framework basis contrast enhancement for uniform with non-uniform back ground imageries utilizing appropriate histogram equalization (CE-VHE), and Contrast enhancement with brightness preservation of low light pictures with the help of combined CLAHE and BPDHE histogram equalization (CE-CLAHE-BPDHE) respectively.

DPAFD-net: A dual-path adaptive fusion dehazing network

Image dehazing is an ill-posed problem that has been extensively studied in recent years. Unfortunately, most existing deep dehazing models have high computational complexity and lack the dynamic adjustment of details, which hinders their application to high-resolution images in computational vision tasks. In this paper, we propose an efficient dual-path adaptive fusion dehazing network (DPAFD-Net) to directly restore a clear image from a hazy input. Moreover, we propose a pure subnetwork with encoder and decoder structures to further extract the structural information and progressively restore the haze-free image. To evaluate the effectiveness of the proposed method, we validate our approach on synthetic and real hazy images, where our method performs favourably against the state-of-the-art dehazing approaches.

Dual‐route synthetic‐to‐real adaption for single image dehazing

AbstractSingle image dehazing in real scenarios is still a particularly challenging task due to the large domain discrepancy between synthetic and real hazy images. To address this problem, the dual‐route domain‐aware adaptation framework with the bridging domain is proposed. First, inspired by the effective multistage strategy in low‐level tasks, the bridging domain is constructed with the color‐preserved adaptive histogram equalization to facilitate the adaptation and improve the performance of real hazy images. Second, the totally shared structure is relaxed and the residual dual‐path domain‐aware modules (RDDM) for synthetic and the bridging domains are proposed, which facilitates extracting the domain‐specific haze information with the different parameters. Third, a half‐cyclic constraint is proposed for the unsupervised hazy images to avoid structure distortion during the unsupervised adversarial training process. Finally, for convenience in the inference stage, feature enhancement modules (FEM) are proposed for the original real hazy images to learn the pre‐process operation in the training stage. Extensive qualitative and quantitative experiments demonstrate that the proposed method significantly improves the dehazing performance on synthetic and real hazy images.

Multi-Scale Attention Feature Enhancement Network for Single Image Dehazing.

Aiming to solve the problem of color distortion and loss of detail information in most dehazing algorithms, an end-to-end image dehazing network based on multi-scale feature enhancement is proposed. Firstly, the feature extraction enhancement module is used to capture the detailed information of hazy images and expand the receptive field. Secondly, the channel attention mechanism and pixel attention mechanism of the feature fusion enhancement module are used to dynamically adjust the weights of different channels and pixels. Thirdly, the context enhancement module is used to enhance the context semantic information, suppress redundant information, and obtain the haze density image with higher detail. Finally, our method removes haze, preserves image color, and ensures image details. The proposed method achieved a PSNR score of 33.74, SSIM scores of 0.9843 and LPIPS distance of 0.0040 on the SOTS-outdoor dataset. Compared with representative dehazing methods, it demonstrates better dehazing performance and proves the advantages of the proposed method on synthetic hazy images. Combined with dehazing experiments on real hazy images, the results show that our method can effectively improve dehazing performance while preserving more image details and achieving color fidelity.

DADRnet: Cross-domain image dehazing via domain adaptation and disentangled representation

Recent years have witnessed remarkable progress of learning-based methods in single image dehazing. Among them, dehazing methods trained on the synthetic images cannot adapt to real hazy ones due to the domain gap, and domain adaptation methods only concentrate on creating a mapping or extracting shared features regardless of representations of deep features. In this paper, we propose an innovative cross-domain dehazing architecture that integrates domain adaptation and disentangled representation. Specifically, we first construct a shared encoder to map synthetic and real hazy images to a latent space and narrow their domain gap at the feature level. Then we utilize a separator to separate the hazy image features into haze-free representations and haze ones. We introduce feature consistency loss and orthogonal loss to further guide the disentanglement process. And then, we utilize a decoder to produce clean images from haze-free features and introduce both supervised and unsupervised losses to guide the training process. Moreover, we also propose to reconstruct the images from both domains by recombining the separated features, which guarantees information completeness. Experiments demonstrate that our method is on par with state-of-the-art methods. Codes are available at https://github.com/lixiaopeng123456/DADRnet.

HDNet: A dual-stream network with progressive fusion for image hazing detection

Image hazing refers to artificially adding haze to a clear image to produce a natural haze effect. However, it might also be used as an image forgery for malicious purposes such as misleading tourists or spreading fake weather information. Up to the present, there are still no works specially designed for fake hazy image detection. Inspired by the fact that almost all artificial hazy images are synthesized based on the atmospheric scattering model (ASM), we observe that though synthetic and real hazy images are very difficult to be differentiated by naked eyes, there still exist some ASM-related imprints in synthesized hazy images. Specifically, the subtle differences between synthetic hazy images and real hazy ones are image saturation and haze distribution. In this work, we propose a dual-stream image hazing detection network, namely HDNet, to distinguish synthetic hazy images from natural hazy ones. One stream explicitly learns features from the image saturation channel, and the other one learns haze distribution features from the dark channel and gray level co-occurrence matrix. Moreover, to fuse well the learned features from two streams, we propose a progressive attention mechanism. A stage-wise learning strategy is exploited for network training, in which the learning process is divided into four sub-tasks that are completed stage-by-stage. Extensive experimental results prove that the proposed HDNet achieves desirable detection accuracy and robustness.

Haze transfer and feature aggregation network for real-world single image dehazing

The absence of ground truth for hazy images in real scenes results in most dehazing models being trained only by synthesized datasets, which is not feasible for real hazy images due to the domain shift. To address this problem, a haze transfer and feature aggregation network, which includes a haze transfer subnetwork and a feature aggregation-based dehazing subnetwork, is developed in this study. In the haze transfer subnetwork, a discriminator group is designed to transfer haze information from real hazy images to clear images. The generated hazy image and its corresponding clear image form a training sample pair to solve the domain shift problem. The deep–shallow feature correlation has been considered by feature aggregation in the dehazing subnetwork. Information related to the haze-free image is highlighted, and the guidance of clear images can be fully used to prevent detailed information loss during feature extraction. The proposed method provides a new aspect of usage for real hazy images during training and improves the scalability of dehazing models. The experimental results show that the dehazing performance of the proposed method outperforms the state-of-the-art dehazing methods compared. The algorithm code is available at https://github.com/lhf12278/HTFA-Net.

Single image dehazing using generative adversarial networks based on an attention mechanism

Most existing image dehazing methods rely on the solution of the atmospheric scattering model or supervised learning based on paired images. However, owing to incomplete prior knowledge and the lack of paired hazy and haze-free images of the same scenes as training samples, their performances for single image dehazing are unsatisfactory. Here, the authors present an unpaired image learning method based on the attention mechanism for single image dehazing problems. The method uses the constraint transfer learning ability and circulatory structure of CycleGAN to carry out an unsupervised image dehazing task for unpaired data. Considering the complexity of the haze distribution in actual imaging and human visual characteristics, the improved channel attention and domain attention mechanisms are integrated into the network to process different features and different regions non-uniformly. The experimental results show that the proposed method achieves good results on both synthetic datasets and real hazy images.

Unsupervised Neural Rendering for Image Hazing.

Image hazing aims to render a hazy image from a given clean one, which could be applied to a variety of practical applications such as gaming, filming, photographic filtering, and image dehazing. To generate plausible haze, we study two less-touched but challenging problems in hazy image rendering, namely, i) how to estimate the transmission map from a single image without auxiliary information, and ii) how to adaptively learn the airlight from exemplars, i.e., unpaired real hazy images. To this end, we propose a neural rendering method for image hazing, dubbed as HazeGEN. To be specific, HazeGEN is a knowledge-driven neural network which estimates the transmission map by leveraging a new prior, i.e., there exists the structure similarity (e.g., contour and luminance) between the transmission map and the input clean image. To adaptively learn the airlight, we build a neural module based on another new prior, i.e., the rendered hazy image and the exemplar are similar in the airlight distribution. To the best of our knowledge, this could be the first attempt to deeply render hazy images in an unsupervised fashion. Compared with existing haze generation methods, HazeGEN renders the hazy images in an unsupervised, learnable, and controllable manner, thus avoiding the labor-intensive efforts in paired data collection and the domain-shift issue in haze generation. Extensive experiments show the promising performance of our method comparing with some baselines in both qualitative and quantitative comparisons. The code is available at https://github.com/XLearning-SCU.

Enhanced densely dehazing network for single image haze removal under railway scenes

Purpose This paper aims to propose an enhanced densely dehazing network to suit railway scenes’ features and improve the visual quality degraded by haze and fog. Design/methodology/approach It is an end-to-end network based on DenseNet. The authors design enhanced dense blocks and fuse them in a pyramid pooling module for visual data’s local and global features. Multiple ablation studies have been conducted to show the effects of each module proposed in this paper. Findings The authors have compared dehazed results on real hazy images and railway hazy images of state-of-the-art dehazing networks with the dehazed results in data quality. Finally, an object-detection test is taken to judge the edge information preservation after haze removal. All results demonstrate that the proposed dehazing network performs better under railway scenes in detail. Originality/value This study provides a new method for image enhancing in the railway monitoring system.

Dense Haze Removal by Nonlinear Transformation

Images captured in hazy or foggy weather conditions, suffer from various problems, such as limited visibility, low contrast, color distortions. These images are used in many computer vision applications, such as video surveillance, transportation, remote sensing. The elimination of the haze effect from these images is essential to ensure the perfect working of these applications. The degradation of a captured image is expressed by the physical model of hazy image formation. The physical model requires the estimation of transmission to restore a haze-free image, which is one of the most important parameters of single image dehazing (SID). Due to the ill-posed nature of SID, lots of priors/assumptions have been used. However, traditional methods fail when these priors do not hold, especially for varying haze concentrations, which lead to many issues such as incomplete haze removal or over enhancement in long-range regions. In this paper, a single image dehazing method based on a superpixel and nonlinear transformation is proposed. The proposed method transforms the minimum filtering on superpixels of a hazy image into the minimum filtering on superpixels of a haze-free image using nonlinear transformation. The nonlinear transformation prevents over enhancement in the long-range regions, while the superpixels reduce the halo artifacts in the dehazed image. The experimental results on challenging real hazy images, dense-hazy images, and synthetic images have proved that a combination of nonlinear transformation and superpixels provide the strength to the proposed method. The obtained dehazed results are evaluated qualitatively and quantitatively and it is found that the proposed method has tremendous performance as compared to state-of-the-art dehazing approaches.

Attentive U-recurrent encoder-decoder network for image dehazing

Haze removal is an important pre-processing step in many computer vision tasks. Convolutional neural networks, especially the U-shaped networks, have shown to be effective in image dehazing. Nevertheless, these networks have three main limitations. First, the relevant haze information, e.g. concentration of haze, is totally ignored. Second, spatial inconsistency and information dilution usually occur when the networks refine the dehazed results with a coarse-to-fine strategy. Third, the receptive field of the network is not large enough to capture structural information. Motivated by these problems, a new attentive U-recurrent encoder-decoder dehazing network is presented, which consists of an attentive recurrent network and a U-recurrent encoder-decoder network. By assuming that haze layers with different depths can be detected by multiple stages, we use an attentive recurrent network to generate the haze attention map for guiding the U-recurrent encoder-decoder network with the concentration of haze to better estimate the clear image. Meanwhile, the features for dehazing are further enhanced and the dehazing results are refined in the U-recurrent encoder-decoder network. This design not only enables spatial consistency but also reduces information dilution with short recurrent pathways. Furthermore, a novel residual pyramid pooling module is also proposed and used in the U-recurrent encoder-decoder network, which provides the network with structural information and with an enlarged receptive field. The experimental results demonstrate that our method outperforms state-of-the-art dehazing algorithms on both synthetic and real hazy images.

URNet: A U-Net based residual network for image dehazing

Low visibility in hazy weather causes the loss of image details in digital images captured by some imaging devices such as monitors. This paper proposes an end-to-end U-Net based residual network (URNet) to improve the visibility of hazy images. The encoder module of URNet uses hybrid convolution combining standard convolution with dilated convolution to expand the receptive field for extracting image features with more details. The URNet embeds several building blocks of ResNet into the junction between the encoder module and the decoder module. This prevents network performance degradation due to the vanishing gradient. After considering large absolute difference on image saturation and value components between hazy images and haze-free images in the HSV color space, the URNet defines a new loss function to better guide the network training. Experimental results on synthetic hazy images and real hazy images show that the URNet significantly improves the image dehazing effect compared to the state-of-the-art methods.

Image dehazing algorithm by combining light field multi-cues and atmospheric scattering model

Image captured in foggy weather often exhibits low contrast and poor image quality, which may have a negative impact on computer vision applications. Aiming at these problems, we propose an image dehazing algorithm by combining light field technology with atmospheric scattering model. Firstly, taking the advantages of capturing multi-view information from light field camera is used to extracting defocus cues and correspondence cues, which are used to estimating the depth information of hazy images, and use the obtained depth information to calculating the scene’s initial transmission. Then use scene depth information to build a new weight function, and combined it with 1-norm context regularization to optimizing the initial transmission map iteratively. Finally, the central perspective image of hazy light field images is dehazed using atmospheric scattering model to obtain the final dehazed images. Experimental results on synthetic hazy images and real hazy images demonstrate that, compared to existing single image dehazing algorithms, the peak signal to noise ratio get 2 dB improvement and the structural similarity raise about 0.04. Moreover, our approach preserves more fine structural information of images and has faithful color fidelity, thus yielding a superior image dehazing result.

A convex single image dehazing model via sparse dark channel prior

In this paper, we present a convex model for single image dehazing via a sparse dark channel prior. Our work is based on an observation that the number of bright pixels is very small in the dark channel of a haze-free image, but significantly increases in that of a hazy image due to the existence of bright atmosphere light. Since the dehazing problem is inherently ambiguous, we first reformulate the degradation model of hazy images into an equivalent form where the transmission and the image variable are decoupled. With the above formulation and observation, we propose the convex model to recover the haze-free image whose dark channel is assumed to be sparse. The proposed objective function consists of four terms: a data-fitting term, an l1 regularization term for the dark channel of the haze-free image, and two total variation regularization terms for both the haze-free image and the transmission map. We develop an efficient alternating direction method of multipliers (ADMM) to tackle the proposed convex model. Extensive experiments on real hazy images illustrate that our method outperforms the state-of-the-art methods.

Multilevel Image Dehazing Algorithm Using Conditional Generative Adversarial Networks

In recent years, the hazy weather in China occurs frequently, and image dehazing has gradually become a research hotspot. To improve the dehazing effect of the hazy images, this paper has proposed a multilevel image dehazing algorithm using conditional generative adversarial networks (CGAN). The hazy image is used to generate the composed image $K$ jointly estimated by a transmission map and atmospheric light value through a generator network, and a dehazed image is calculated through an improved atmospheric scattering model. The generator network and the joint discriminator network are subjected to adversarial training and reconstruction constraints. The experimental results show that the proposed method achieved good dehazing effect in synthetic hazy images and real hazy images, and is ahead of other advanced dehazing methods in subjective evaluation and objective evaluation.

Quality Evaluation of Image Dehazing Methods Using Synthetic Hazy Images

To enhance the visibility and usability of images captured in hazy conditions, many image dehazing algorithms (DHAs) have been proposed. With so many image DHAs, there is a need to evaluate and compare these DHAs. Due to the lack of the reference haze-free images, DHAs are generally evaluated qualitatively using real hazy images. But it is possible to perform quantitative evaluation using synthetic hazy images since the reference haze-free images are available and full-reference (FR) image quality assessment (IQA) measures can be utilized. In this paper, we follow this strategy and study DHA evaluation using synthetic hazy images systematically. We first build a synthetic haze removing quality (SHRQ) database. It consists of two subsets: regular and aerial image subsets, which include 360 and 240 dehazed images created from 45 and 30 synthetic hazy images using 8 DHAs, respectively. Since aerial imaging is an important application area of dehazing, we create an aerial image subset specifically. We then carry out subjective quality evaluation study on these two subsets. We observe that taking DHA evaluation as an exact FR IQA process is questionable, and the state-of-the-art FR IQA measures are not effective for DHA evaluation. Thus, we propose a DHA quality evaluation method by integrating some dehazing-relevant features, including image structure recovering, color rendition, and over-enhancement of low-contrast areas. The proposed method works for both types of images, but we further improve it for aerial images by incorporating its specific characteristics. Experimental results on two subsets of the SHRQ database validate the effectiveness of the proposed measures.

Objective Quality Evaluation of Dehazed Images

Vision-based intelligent systems like automatic driving or driving assistance can be improved by enhancing the visibility of the scenes captured in bad weather conditions. In particular, many image dehazing algorithms (DHAs) have been proposed to facilitate such applications in hazy weather. Contrary to the substantial progress of DHA developing, the quality evaluation of DHAs falls behind. Generally, DHAs can be evaluated qualitatively by human subjects or quantitatively by objective quality measures. Compared with the subjective evaluation which is time consuming and difficult to apply, objective measures with quantitative results are more needed in practical systems. But in the literature, very few measures are widely utilized, and even less measures correlate well with the overall dehazing quality (DHQ). In this paper, we study the DHQ evaluation using real hazy images systematically. We first construct a DHQ database, which is the largest of its kind so far and includes 1750 dehazed images generated from 250 real hazy images of various haze densities using seven representative DHAs. A subjective quality evaluation study is subsequently conducted on the DHQ database. Then, we propose an objective DHQ index (DHQI) by extracting and fusing three groups of features, including: 1) haze-removing features; 2) structure-preserving features; and 3) over-enhancement features, which have captured the most key aspects of dehazing. DHQI can be utilized to evaluate DHAs or optimize practical dehazing systems. Validations on the constructed DHQ database and three other databases with synthetic haze have verified the effectiveness of DHQI. Finally, we give an overview of the current DHA quality evaluation strategies, discuss their merits and demerits, and give some suggestions on systematic DHA quality evaluation. The DHQ database and the code of DHQI will be released to facilitate further research.