Dehazing Methods Research Articles

Haze removal network using unified function for image dehazing

The atmospheric scattering model includes two crucial parameters for dehazing: global atmospheric light and the transmission map. Most previous dehazing methods need to obtain these two parameters separately, which makes dehazing a difficult and ill-posed problem. Here, a new unified function that includes both the crucial parameters for haze removal is proposed. Then the haze removal network, which now needs to learn only one function during training, is proposed. Image dehazing can be performed as a simple addition of the haze removal function with the input hazy image. Experimental results show that the proposed method gives better subjective results for indoor and outdoor synthesized images as well as natural images compared to previous state-of-the-art methods. Quantitative evaluation results show that the proposed haze removal network gives improved objective results compared with the same previous methods.

A modular architecture for high resolution image dehazing

Haze is an atmospheric phenomenon which diminishes visibility in outdoor images. Algorithms based on dark channel prior (DCP) and haze line prior are found to be effective for dehazing images. These two methods make use of the Laplacian matrix, which is computationally complex, memory intensive and slow, thus making it impossible to use them on high-resolution (large) images. Multiple strategies have been suggested in the literature to speed up dehazing process by avoiding the Laplacian matrix, but these methods compromise on the quality of dehazing. We propose an effective modular method which divides the input image into blocks and processes each block independently. This makes it possible to use our method for dehazing large images retaining Laplacian matting and thus ensuring the output image quality. This division results in the possibility of assuming local values of atmospheric light. We show that this approach results in better dehazing in the local regions. The effectiveness of the proposed modular architecture is tested also on a learning based method. The output of the modular method is compared with those of different state-of-the-art dehazing methods for multiple quality parameters. Toward this, we have created a dataset of hazy natural outdoor images of large size.

SIDE-A Unified Framework for Simultaneously Dehazing and Enhancement of Nighttime Hazy Images.

Single image dehazing is a difficult problem because of its ill-posed nature. Increasing attention has been paid recently as its high potential applications in many visual tasks. Although single image dehazing has made remarkable progress in recent years, they are mainly designed for haze removal in daytime. In nighttime, dehazing is more challenging where most daytime dehazing methods become invalid due to multiple scattering phenomena, and non-uniformly distributed dim ambient illumination. While a few approaches have been proposed for nighttime image dehazing, low ambient light is actually ignored. In this paper, we propose a novel unified nighttime hazy image enhancement framework to address the problems of both haze removal and illumination enhancement simultaneously. Specifically, both halo artifacts caused by multiple scattering and non-uniformly distributed ambient illumination existing in low-light hazy conditions are considered for the first time in our approach. More importantly, most current daytime dehazing methods can be effectively incorporated into nighttime dehazing task based on our framework. Firstly, we decompose the observed hazy image into a halo layer and a scene layer to remove the influence of multiple scattering. After that, we estimate the spatially varying ambient illumination based on the Retinex theory. We then employ the classic daytime dehazing methods to recover the scene radiance. Finally, we generate the dehazing result by combining the adjusted ambient illumination and the scene radiance. Compared with various daytime dehazing methods and the state-of-the-art nighttime dehazing methods, both quantitative and qualitative experimental results on both real-world and synthetic hazy image datasets demonstrate the superiority of our framework in terms of halo mitigation, visibility improvement and color preservation.

CSIDNet: Compact single image dehazing network for outdoor scene enhancement

This paper proposes a novel deep learning-based single image dehazing network named as Compact Single Image Dehazing Network (CSIDNet) for outdoor scene enhancement. CSIDNet directly outputs a haze-free image from the given hazy input. The remarkable features of CSIDNet are that it has been designed only with three convolutional layers and it requires lesser number of images for training without diminishing the performance in comparison to the other commonly observed deep learning-based dehazing models. The performance of CSIDNet has been analyzed on natural hazy scene images and REalistic Single Image DEhazing (RESIDE) dataset. RESIDE dataset consists of Outdoor Training Set (OTS), Synthetic Objective Testing Set (SOTS), and real-world synthetic hazy images from Hybrid Subjective Testing Set (HSTS). The performance metrics used for comparison are Peak Signal to Noise Ratio (PSNR) and Structural SIMilarity (SSIM) index. The experimental results obtained using CSIDNet outperform several well known state-of-the-art dehazing methods in terms of PSNR and SSIM on images of SOTS and HSTS from RESIDE dataset. Additionally, the visual comparison shows that the dehazed images obtained using CSIDNet are more appealing with better edge preservation. Since the proposed network requires minimal resources and is faster to train along with lesser run-time, it is more practical and feasible for real-time applications.

Aerial image dehazing using a deep convolutional autoencoder

Aerial images and videos are extensively used for object detection and target tracking. However, due to the presence of thin clouds, haze or smoke from buildings, the processing of aerial data can be challenging. Existing single-image dehazing methods that work on ground-to-ground images, do not perform well on aerial images. Moreover, current dehazing methods are not capable for real-time processing. In this paper, a new end-to-end aerial image dehazing method using a deep convolutional autoencoder is proposed. Using the convolutional autoencoder, the dehazing problem is divided into two parts, namely, encoder, which aims extract important features to dehaze hazy regions and decoder, which aims to reconstruct the dehazed image using the down-sampled image received from the encoder. In this proposed method, we also exploit the superpixels in two different scales to generate synthetic thin cloud data to train our network. Since this network is trained in an end-to-end manner, in the test phase, for each input hazy aerial image, the proposed algorithm outputs a dehazed version without requiring any other information such as transmission map or atmospheric light value. With the proposed method, hazy regions are dehazed and objects within hazy regions become more visible while the contrast of non-hazy regions is increased. Experimental results on synthetic and real hazy aerial images demonstrate the superiority of the proposed method compared to existing dehazing methods in terms of quality and speed.

Improvement of underwater colour correction using standard deviation ratio

This Letter presents a simple and new colour correction algorithm for the underwater image. The proposed method corrects the pixels of the weak colour channel based on the ratio of the standard deviation of the weak colour channel to that of the green channel. The corrected image based on the proposed colour balanced algorithm barely produces a reddish artefact and can be applied directly to the existing dehazing approaches without any processing. The experimental results show that the proposed algorithm achieves good colour correction performance, as well as reasonable underwater image enhancement when employed with the existing dehazing methods.

Deep joint neural model for single image haze removal and color correction

The quality of an image affects the performance of computer vision applications. The presence of haze often greatly depreciates the visual effect of images. It is a traditional and critical vision challenge to remove haze from a single image. This paper proposes a trainable end-to-end de-hazing connectionist model with a special design. First, feature learning is conducted using hierarchical convolutional layers with nested structures. Cascaded haze-relevant tasks are then sequentially performed via a physics-driven sub-network. In particular, to break the assumption of a homogeneous atmosphere, a branch of the sub-network estimates the scattering factor in the form of a two-dimensional tensor. Finally, a chromatic adaptation layer is proposed for color adjustment, which is often neglected in existing de-hazing methods. In addition, we integrate different training criteria based on the characteristics of the haze-relevant variables in our model. For a fully actionable optimization, an asynchronous learning paradigm is designed for the fusion of different de-hazing tasks, and the joint model is further facilitated by a cyclic restoration. The effectiveness of the proposed de-hazing model was verified via extensive experiments, and most results of our method are remarkable.

FD-GAN: Generative Adversarial Networks with Fusion-Discriminator for Single Image Dehazing

Recently, convolutional neural networks (CNNs) have achieved great improvements in single image dehazing and attained much attention in research. Most existing learning-based dehazing methods are not fully end-to-end, which still follow the traditional dehazing procedure: first estimate the medium transmission and the atmospheric light, then recover the haze-free image based on the atmospheric scattering model. However, in practice, due to lack of priors and constraints, it is hard to precisely estimate these intermediate parameters. Inaccurate estimation further degrades the performance of dehazing, resulting in artifacts, color distortion and insufficient haze removal. To address this, we propose a fully end-to-end Generative Adversarial Networks with Fusion-discriminator (FD-GAN) for image dehazing. With the proposed Fusion-discriminator which takes frequency information as additional priors, our model can generator more natural and realistic dehazed images with less color distortion and fewer artifacts. Moreover, we synthesize a large-scale training dataset including various indoor and outdoor hazy images to boost the performance and we reveal that for learning-based dehazing methods, the performance is strictly influenced by the training data. Experiments have shown that our method reaches state-of-the-art performance on both public synthetic datasets and real-world images with more visually pleasing dehazed results.

An end-to-end dehazing network with transitional convolution layer

Image dehazing is a challenging task of restoring a clear image from a haze-polluted image. However, most popular dehazing methods have color shift and overexposure problems owing to the inaccurate estimation of the transmission and atmospheric light in the atmospheric scattering model. In view of the existing problems, this paper proposes a simple but effective dehazing network by learning the residual image between the hazy image and haze-free image. A novel transitional convolution structure is constructed in this network, which contributes to utilizing shallow structure information to enhance the final residual image. Furthermore, to provide images with more realistic color information after dehazing, a novel color difference loss based on CIEDE2000 is designed as one term of the total loss function. In addition, for some real-world images that affect practical applications, a brightness enhancement module is also introduced to restore the luminance of the images. Experiments on synthetic datasets and real-world images demonstrate that the proposed method has clear advantages in both subjective and objective evaluation indicators compared to several existing advanced algorithms.

Generalized Polarimetric Dehazing Method Based on Low-Pass Filtering in Frequency Domain.

Polarimetric dehazing methods can significantly enhance the quality of hazy images. However, current methods are not robust enough under different imaging conditions. In this paper, we propose a generalized polarimetric dehazing method based on low-pass filtering in the frequency domain. This method can accurately estimate the polarized state of the scattering light automatically without adjusting bias parameters. Experimental results show the effectiveness and robustness of our proposed method in different hazy weather and scattering underwater environments with different densities. Furthermore, computational efficiency is enhanced more than 70% compared to the polarimetric dehazing method we proposed previously.

Adaptive single image dehazing method based on support vector machine

A dehazing method often only shows good results when processing the image for a certain haze concentration. So an adaptive hazy image dehazing method based on SVM is proposed. The innovation points are as follows: Firstly, combining the characteristics of the degraded images of haze weather, the dark channel histogram and texture features of the input images are extracted to form the feature vectors. These are trained by supervised learning through SVM algorithm to realize automatic binary classification of images; Secondly, the defined dehazing methods are called to process the classified result as a hazy image and the same quality evaluation indexes are used to evaluate each image output by different dehazing methods. Then, it outputs the highest evaluation image after haze removal. Finally, the output image is classified again by SVM until the image reaches the clearest it can be. The experimental results show that the proposed algorithm exhibits good contrast, brightness and color saturation from the visual effect. Also the scene adaptability and robustness of the algorithm are improved.

A Survey on Video Dehazing Using Deep Learning

With the fast development of object recognition and detection in autonomous driving and video monitoring, image with haze or raindrop can affect the result a lot. As deep learning develops, the hazed and raindrop image can lower the accuracy of object recognition and detection significantly. While hazed images cannot be managed using other image refining process. Since the noise in hazed image is signal-dependent. The object degradation in hazed image is related to object depth. So, the dehazing process depends on the input image. This paper provides a survey on single image and video dehazing methods, from end-to-end system to distributed system. General methods based on deep learning of state-of-art papers from 2010 to 2018 are summarized and compared, accompanied with their datasets of the current progress in this field. The application of these methods and relationship between these methos are also discussed in this paper.

Image Dehazing Based on (CMTnet) Cascaded Multi-scale Convolutional Neural Networks and Efficient Light Estimation Algorithm

Image dehazing plays a pivotal role in numerous computer vision applications such as object recognition, surveillance systems, and security systems, where it can be considered as an introductory stage. Recently, many proposed learning-based works address this significant task; however, most of them neglect the atmospheric light estimation and fail to produce accurate transmission maps. To address such a problem, in this paper, we propose a two-stage dehazing system. The first stage presents an accurate atmospheric light algorithm labeled “A-Est” that employs hazy image blurriness and quadtree decomposition. Te second stage represents a cascaded multi-scale CNN model called CMT n e t that consists of two subnetworks, one for calculating rough transmission maps (CMCNN t r ) and the other for its refinement (CMCNN t ). Each subnetwork is composed of three-layer D-units (D indicates dense). Experimental analysis and comparisons with state-of-the-art dehazing methods revealed that the proposed system can estimate AL and t efficiently and accurately by achieving high-quality dehazing results and outperforms state-of-the-art comparative methods according to SSIM and MSE values, where our proposed achieves the best scores of both (91% average SSIM and 0.068 average MSE).

Improvement of dehazing algorithm based on dark channel priori theory

The dehazing algorithm based on dark channel priori theory is an effective method for dehazing of single image. However, it still has shortcomings in some of special cases. In this work, a method is proposed to impair these disadvantages. By considering local smoothness of the sky region, pixels in the bright regions (such as light source, white object, etc.) of the haze image are first excluded from being mistaken as reference pixels, so that the estimation of subsequent parameters is more accurate. Meanwhile, the pixels of the most occurrences are selected to replace the ones in the neighborhood centered on the edge pixels, and the operation is carried out pixel by pixel along the edge contour in replace of the minimum filter. This makes the estimation of transmission near the edge in better agreement with the actual situation, and can effectively avoid the “white haze” which is an artifact caused by the minimum filtering in the areas with the depth of field changing dramatically. Compared with the previous methods, the proposed algorithm in this paper can restore images more clearly, with more image edge details retained, and effectively improve visual effect of the scene in haze weather.

Accurate estimation of transmission maps for image restoration based on polarimetric parameters and average intensity

Haze limits visibility and degrades image contrast, and images lose color fidelity owing to the scattering and absorption of light caused by particles. Consequently, removing haze is critical for enhancing the image visibility and quality, with estimation of airlight and transmission map being critical steps. We propose a simple and convenient improved scheme, which is polarization-based method for dehazing. The method uses four polarization orientations (0°, 45°, 90°, 135°) to measure the Stokes vector, and obtained the degree of polarization (DOP) of the airlight by angle of polarization (AOP) in three channels. Subsequently, estimating the airlight at infinity by average total intensity and re-constructing a matrix to estimate the transmission maps. Experimental results demonstrate that a high-quality transmission map can be obtained, and this improved method can restore outdoor scene haze images with high quality and contrast. In addition, this method also proved to be useful for enhancing color saturation and restoring color fidelity.

Task-Oriented Network for Image Dehazing.

Haze interferes the transmission of scene radiation and significantly degrades color and details of outdoor images. Existing deep neural networks-based image dehazing algorithms usually use some common networks. The network design does not model the image formation of haze process well, which accordingly leads to dehazed images containing artifacts and haze residuals in some special scenes. In this paper, we propose a task-oriented network for image dehazing, where the network design is motivated by the image formation of haze process. The task-oriented network involves a hybrid network containing an encoder and decoder network and a spatially variant recurrent neural network which is derived from the hazy process. In addition, we develop a multi-stage dehazing algorithm to further improve the accuracy by filtering haze residuals in a step-bystep fashion. To constrain the proposed network, we develop a dual composition loss, content-based pixel-wise loss and total variation constraint. We train the proposed network in an end-to-end manner and analyze its effect on image dehazing. Experimental results demonstrate that the proposed algorithm achieves favorable performance against state-of-the-art dehazing methods.

Lower bound on Transmission using Non-Linear Bounding Function in Single Image Dehazing.

The visibility of an image captured in poor weather (such as haze, fog, mist, smog) degrades due to scattering of light by atmospheric particles. Single image dehazing (SID) methods are used to restore visibility from a single hazy image. The SID is a challenging problem due to its ill-posed nature. Typically, the atmospheric scattering model (ATSM) is used to solve SID problem. The transmission and atmospheric light are two prime parameters of ATSM. The accuracy and effectiveness of SID depends on accurate value of transmission and atmospheric light. The proposed method translates transmission estimation problem into estimation of the difference between minimum color channel of hazy and haze-free image. The translated problem presents a lower bound on transmission and is used to minimize reconstruction error in dehazing. The lower bound depends upon the bounding function (BF) and a quality control parameter. A non-linear model is then proposed to estimate BF for accurate estimation of transmission. The proposed quality control parameter can be utilized to tune the effect of dehazing. The accuracy obtained by the proposed method for transmission is compared with state of the art dehazing methods. Visual comparison of dehazed images and objective evaluation further validates the effectiveness of the proposed method.

Integrating Haze Density Features for Fast Nighttime Image Dehazing

To date, much progress has been achieved on daytime image dehazing, yet the nighttime dehazing problem is still not well addressed. Different from the imaging conditions in the daytime, the ambient illumination in the nighttime hazy scene is usually not globally isotropic due to the non-uniform incident lights from multiple artificial light sources. Currently, almost all the existing nighttime dehazing methods use a certain kind of image priors, whereby these spatial filtering based priors are not widely applicable in nighttime hazy scenes. For example, the maximum reflectance prior (MRP) cannot handle the dark regions well and the dark channel prior (DCP) is not valid in the light source regions. In this paper, we propose an efficient and fast method for nighttime image dehazing. By exploring the visual properties of hazy images, we construct an effective linear model to build the connection between the transmission and multiple haze-relevant features. Towards solving this model, a data-driven approach is adopted to learn the unknown coefficients. Operating on the pixel level, this novel approach requires no further refinement of transmission map as used in those prior-based methods. In addition, aiming at solving the problem of halo effect around the light sources caused by MRP, we introduce a color-dependant MRP method for color correction. We demonstrate the effectiveness of our method on a number of experiments compared to the state-of-the-art nighttime dehazing methods.

水下和微光图像的去雾及增强方法

水下和微光图像的去雾及增强方法

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.