Dehazing Model Research Articles

Adaptive Multi-Feature Attention Network for Image Dehazing

Currently, deep-learning-based image dehazing methods occupy a dominant position in image dehazing applications. Although many complicated dehazing models have achieved competitive dehazing performance, effective methods for extracting useful features are still under-researched. Thus, an adaptive multi-feature attention network (AMFAN) consisting of the point-weighted attention (PWA) mechanism and the multi-layer feature fusion (AMLFF) is presented in this paper. We start by enhancing pixel-level attention for each feature map. Specifically, we design a PWA block, which aggregates global and local information of the feature map. We also employ PWA to make the model adaptively focus on significant channels/regions. Then, we design a feature fusion block (FFB), which can accomplish feature-level fusion by exploiting a PWA block. The FFB and PWA constitute our AMLFF. We design an AMLFF, which can integrate three different levels of feature maps to effectively balance the weights of the inputs to the encoder and decoder. We also utilize the contrastive loss function to train the dehazing network so that the recovered image is far from the negative sample and close to the positive sample. Experimental results on both synthetic and real-world images demonstrate that this dehazing approach surpasses numerous other advanced techniques, both visually and quantitatively, showcasing its superiority in image dehazing.

HEDehazeNet: Unpaired image dehazing via enhanced haze generation

Unpaired image dehazing models based on Cycle-Consistent Adversarial Networks (CycleGAN) typically consist of two cycle branches: dehazing-rehazing branch and hazing-dehazing branch. In these two branches, there is an asymmetry of information in the mutual transformation process between haze images and haze-free images. Previous models tended to focus more on the transformation process from haze images to haze-free images within the dehazing-rehazing branch, overlooking the provision of effective information for the formation of haze images in the hazing-dehazing branch. This oversight results in the production of haze patterns that are both monotonous and simplistic, ultimately impeding the overall performance and generalization capabilities of dehazing networks. In light of this, this paper proposes a novel model called HEDehazeNet (Dehazing Net based on Haze Generation Enhancement), which provides crucial information for the generation process of haze images through a dedicated haze generation enhancement module. This module is capable of producing three distinct modes of transmission maps - random transmission map, simulated transmission map, and mixed transmission maps combining both. Employing these transmission maps to generate hazing images with varying density and patterns provides the dehazing network with a more diverse and dynamically complex set of training samples, thereby enhancing its capacity to handle intricate scenes. Additionally, we made minor modifications to the U-Net, replacing residual blocks with multi-scale parallel convolutional blocks and channel self-attention, to further enhance the network's performance. Experiments were conducted on both synthetic and real-world datasets, substantiating the superiority of HEDehazeNet over the current state-of-the-art unpaired dehazing models.

An Efficient Attentional Image Dehazing Deep Network Using Two Color Space (ADMC2-net).

Image dehazing has become a crucial prerequisite for most outdoor computer applications. The majority of existing dehazing models can achieve the haze removal problem. However, they fail to preserve colors and fine details. Addressing this problem, we introduce a novel high-performing attention-based dehazing model (ADMC2-net)that successfully incorporates both RGB and HSV color spaces to maintain color properties. This model consists of two parallel densely connected sub-models (RGB and HSV) followed by a new efficient attention module. This attention module comprises pixel-attention and channel-attention mechanisms to get more haze-relevant features. Experimental results analyses can validate that our proposed model (ADMC2-net) can achieve superior results on synthetic and real-world datasets and outperform most of state-of-the-art methods.

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.

RepDehazeNet: Dual subnets image dehazing network based on structural re-parameterization

In recent times, there has been notable and swift advancement in the field of image dehazing. Several deep learning techniques have demonstrated remarkable proficiency in resolving homogeneous dehazing issues. Nonetheless, the current dehazing approaches are generally formulated to deal with homogeneous haze, which is often undermined in real-world scenarios due to the uncertain haze dispersion. In this paper, we propose a dehazing model named RepDehazeNet by combining a structurally Reparameterization Encoder-Decoder subnet and a Full-Resolution Attention subnet. To be specific, the structural reparameterization idea is introduced into the encoder–decoder subnet to strengthen the feature extraction of dehazed images and improve the feature extraction speed. RepDehazeNet is compared with seven SOTA models on different datasets in terms of PSNR, SSIM, parameter quantity, and inference time. Compared to the DW-GAN model, the proposed RepDehazeNet model reduces the number of parameters by 2.7 million, and improves the inference speed by 90.3%, while achieving a higher PSNR of 0.5 dB on the NH-Haze2021 dataset. The experimental results demonstrate that the proposed RepDehazeNet model can effectively improve the real-time performance, accuracy of dehazing synthesized and nonhomogeneous haze images.

Spectral Dual-Channel Encoding for Image Dehazing

In recent years, deep learning-based dehazing models have presented a momentum of dramatic growth. Unfortunately, most deep learning-based approaches heavily rely on synthetically hazed images for model training, which makes these methods brittle to restore hazy images taken from real-world scenes, due to the sample distribution discrepancy between synthetic and realistic images. Although some attempts have been made to overcome this difficulty by augmenting image spatial features with spectral features, the power of the spectral features still remains underutilized. In this paper, we propose the Spectral Dual-Channel Encoding (SDCE) framework for high-quality image dehazing, by unleashing the power of spectral feature encoding. We argue that hazes impose more adverse impacts on high-frequency image features (e.g., outlines and textures) than low-frequency features (e.g., colors), with theoretical and empirical justifications. To better restore hazed high- and low-frequency features, we decompose the hazed images into high- and low-frequency feature components with spectral dual-channel encoding and respectively design effective neural network architectures to recover hazed images on the two feature components. To be specific, we recover the low-frequency feature components with an encoder-decoder, while we specially design a high-frequency aggregation component (HFAC) to recover hazed images on high-frequency feature components, by referring to neighboring feature distributions. We conduct extensive experiments on four real-world image dehazing benchmarks. The experimental results show that our proposed SDCE framework outperforms the state-of-the-art baselines significantly, with an average 4.4% improvement in PSNR and an average 7.7% gain in SSIM.

DFFA-Net: A Differential Convolutional Neural Network for Underwater Optical Image Dehazing

This paper proposes DFFA-Net, a novel differential convolutional neural network designed for underwater optical image dehazing. DFFA-Net is obtained by deeply analyzing the factors that affect the quality of underwater images and combining the underwater light propagation characteristics. DFFA-Net introduces a channel differential module that captures the mutual information between the green and blue channels with respect to the red channel. Additionally, a loss function sensitive to RGB color channels is introduced. Experimental results demonstrate that DFFA-Net achieves state-of-the-art performance in terms of quantitative metrics for single-image dehazing within convolutional neural network-based dehazing models. On the widely-used underwater Underwater Image Enhancement Benchmark (UIEB) image dehazing dataset, DFFA-Net achieves a peak signal-to-noise ratio (PSNR) of 24.2631 and a structural similarity index (SSIM) score of 0.9153. Further, we have deployed DFFA-Net on a self-developed Remotely Operated Vehicle (ROV). In a swimming pool environment, DFFA-Net can process hazy images in real time, providing better visual feedback to the operator. The source code has been open sourced.

Using Haze Level Estimation in Data Cleaning for Supervised Deep Image Dehazing Models

Recently, supervised deep learning methods have been widely used for image haze removal. These methods rely on training data that are assumed to be appropriate. However, this assumption may not always be true. We observe that some data may contain hazy ground truth (GT) images. This can lead to supervised deep image dehazing (SDID) models learning inappropriate mapping between hazy images and GT images, which negatively affects the dehazing performance. To address this problem, two difficulties must be solved. One is to estimate the haze level in an image, and the other is to develop a haze level indicator to discriminate clear and hazy images. To this end, we proposed a haze level estimation (HLE) scheme based on dark channel prior and a haze level indicator accordingly for training data cleaning, i.e., to exclude image pairs with hazy GT images in the data set. With the data cleaning by the HLE, we introduced an SDID framework to avoid inappropriate learning and thus improve the dehazing performance. To verify the framework, using the RESIDE data set, experiments were conducted with three types of SDID models, i.e., GCAN, REFN and cGAN. The results show that our method can significantly improve the dehazing performance of the three SDID models. Subjectively, the proposed method generally provides better visual quality. Objectively, our method, using fewer training image pairs, was capable of improving PSNR in the GCAN, REFN, and cGAN models by 3.10 dB, 5.74 dB, and 6.44 dB, respectively. Furthermore, our method was evaluated using a real-world data set, KeDeMa. The results indicate that the better visual quality of the dehazed images is generally for models with the proposed data cleaning scheme. The results demonstrate that the proposed method effectively and efficiently enhances the dehazing performance in the given examples. The practical significance of this research is to provide an easy but effective way, that is, the proposed data cleaning scheme, to improve the performance of SDID models.

Mueller transform matrix neural network for underwater polarimetric dehazing imaging.

Polarization dehazing imaging has been used to restore images degraded by scattering media, particularly in turbid water environments. While learning-based approaches have shown promise in improving the performance of underwater polarimetric dehazing, most current networks rely heavily on data-driven techniques without consideration of physics principles or real physical processes. This work proposes, what we believe to be, a novel Mueller transform matrix network (MTM-Net) for underwater polarimetric image recovery that considers the physical dehazing model adopting the Mueller matrix method, significantly improving the recovery performance. The network is trained with a loss function that combines content and pixel losses to facilitate detail recovery, and is sped up with the inverse residuals and channel attention structure without decreasing image recovery quality. A series of ablation experiment results and comparative tests confirm the performance of this method with a better recovery effect than other methods. These results provide deeper understanding of underwater polarimetric dehazing imaging and further expand the functionality of polarimetric dehazing method.

Aethra-net: Single image and video dehazing using autoencoder

A fast and efficient video dehazing system with low computational complexity has a huge demand among drivers during hazy winter nights. There are only a few video dehazing models that exist in literature. Video dehazing requires the sequential extraction and processing of frames. The processed frames must be restored in the same sequence as the original video. However, the existing video dehazing algorithms suffer from color distortion due to the continuous processing of frames. They are not suitable for videos with dense haze. Furthermore, some dehazing systems require hardware, whereas the proposed model is completely software-based to reduce the computational costs. In this paper, an image and video dehazing system called Aethra-Net is developed. A gush enhancer-based autoencoder is modified to obtain the transmission map. The structure of gush enhancement module resembles the processing of light entering the human eye from different paths. The multiple blocks of Resnet-101 layers are employed to overcome vanishing gradient problem. The vessel enhancement filter is also incorporated to enhance the performance of the proposed system. The proposed model has a susceptibility to compute the dehazed images effectively. The proposed model is evaluated on various benchmark datasets and compared with the existing dehazing techniques. Experimental results reveal that the performance of Aethra-Net is found superior as compared to the existing dehazing models.

Detection of Targets in Road Scene Images Enhanced Using Conditional GAN-Based Dehazing Model

Object detection is a classic image processing problem. For instance, in autonomous driving applications, targets such as cars and pedestrians are detected in the road scene video. Many image-based object detection methods utilizing hand-crafted features have been proposed. Recently, more research has adopted a deep learning approach. Object detectors rely on useful features, such as the object’s boundary, which are extracted via analyzing the image pixels. However, the images captured, for instance, in an outdoor environment, may be degraded due to bad weather such as haze and fog. One possible remedy is to recover the image radiance through the use of a pre-processing method such as image dehazing. We propose a dehazing model for image enhancement. The framework was based on the conditional generative adversarial network (cGAN). Our proposed model was improved with two modifications. Various image dehazing datasets were employed for comparative analysis. Our proposed model outperformed other hand-crafted and deep learning-based image dehazing methods by 2dB or more in PSNR. Moreover, we utilized the dehazed images for target detection using the object detector YOLO. In the experimentations, images were degraded by two weather conditions—rain and fog. We demonstrated that the objects detected in images enhanced by our proposed dehazing model were significantly improved over those detected in the degraded images.

Towards generalization on real domain for single image dehazing via meta-learning

Learning-based image dehazing methods are essential to assist autonomous systems in enhancing reliability. Due to the domain gap between synthetic and real domains, the internal information learned from synthesized images is usually sub-optimal for real domains. Thus, the dehazing models trained on synthetic images often suffer from severe performance drops on real-world samples. Driven by the ability to explore internal information from a few unseen-domain samples, meta-learning is commonly adopted to address this issue via test-time training, which is hyperparameter-sensitive and time-consuming. In contrast, we present a domain generalization framework based on meta-learning to dig out representative and discriminative internal properties of real hazy domains without test-time training. To obtain representative domain-specific information, we attach two entities termed adaptation network and distance-aware aggregator to our dehazing network. The adaptation network assists in distilling domain-relevant information from a few hazy samples and caching it into a collection of features. The distance-aware aggregator strives to summarize the generated features and filter out misleading information for more representative internal properties. To enhance the discrimination of distilled internal information, we present a novel loss function called domain-relevant contrastive regularization. It encourages the similarity of the internal features generated from the same domain and the distinction of the features from diverse domains. The generated representative and discriminative features are regarded as some external variables of our dehazing network to regress a particular function for a given domain. The extensive experiments on real hazy datasets validate that our proposed method has superior generalization ability than the state-of-the-art competitors.

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

Captured images of outdoor scenes usually exhibit low visibility in cases of severe haze, which interferes with optical imaging and degrades image quality. Most of the existing methods solve the single-image dehazing problem by applying supervised training on paired images; however, in practice, the pairing of real-world images is not viable. Additionally, the processing speed of individual dehazing models is important in practical applications. In this study, a novel unsupervised single image dehazing network (USID-Net) based on disentangled representations without paired training images is explored. Furthermore, considering the trade-off between performance and memory storage, a compact multi-scale feature attention (MFA) module is developed, integrating multi-scale feature representation and attention mechanism to facilitate feature representation. To effectively extract haze information, a mechanism referred to as OctEncoder is designed to include multi-frequency representations that can capture more global information. Extensive experiments show that USID-Net achieves competitive dehazing results and a relatively high processing speed compared to state-of-the-art methods.

A dehazing method for flight view images based on transformer and physical priori.

Aiming at the problems of local dehazing distortion and incomplete global dehazing of existing algorithms in real airborne cockpit environments, a two-stage dehazing method PhysiFormer combining physical a priori with a Transformer oriented flight perspective was proposed. The first stage used synthetic pairwise data to pre-train the dehazing model. First, a pyramid pooling module (PPM) was introduced in the Transformer for multiscale feature extraction to solve the problem of poor recovery of local details, then a global context fusion mechanism was used to enable the model to better perceive global information. Finally, considering that combining the physical a priori needs to rely on the estimation of the atmosphere light, an encoding-decoding structure based on the residual blocks was used to estimate the atmosphere light, which was then used for dehazing through the atmospheric scattering model for dehazing. The second stage used real images combined with physical priori to optimize the model to better fit the real airborne environment. The experimental results show that the proposed method has better naturalness image quality evaluator (NIQE) and blind/referenceless image spatial quality evaluator (BRISQUE) indexes and exhibits the best dehazing visual effect in the tests of dense haze, non-uniform haze and real haze images, which effectively improves the problems of color distortion and haze residue.

Enhancement of Underwater Images by CNN-Based Color Balance and Dehazing

Convolutional neural networks (CNNs) are employed to achieve the color balance and dehazing of degraded underwater images. In the module of color balance, an underwater generative adversarial network (UGAN) is constructed. The mapping relationship between underwater images with color deviation and clean underwater images is learned. In the module of clarity improvement, an all-in-one dehazing model is proposed in which a comprehensive index is introduced and estimated by deep CNN. The third module to enhance underwater images adopts an adaptive contrast improvement method by fusing global and local histogram information. Combined with several underwater image datasets, the proposed enhancement method based on the three modules is evaluated, both by subjective visual effects and quantitative evaluation metrics. To demonstrate the advantages of the proposed method, several commonly used underwater image enhancement algorithms are compared. The comparison results indicate that the proposed method gains better enhancement effects for underwater images in different scenes than the other enhancement algorithms, since it can significantly diminish the color deviation, blur, and low contrast in degraded underwater images.

Uncertainty-Driven Dehazing Network

Deep learning has made remarkable achievements for single image haze removal. However, existing deep dehazing models only give deterministic results without discussing the uncertainty of them. There exist two types of uncertainty in the dehazing models: aleatoric uncertainty that comes from noise inherent in the observations and epistemic uncertainty that accounts for uncertainty in the model. In this paper, we propose a novel uncertainty-driven dehazing network (UDN) that improves the dehazing results by exploiting the relationship between the uncertain and confident representations. We first introduce an Uncertainty Estimation Block (UEB) to predict the aleatoric and epistemic uncertainty together. Then, we propose an Uncertainty-aware Feature Modulation (UFM) block to adaptively enhance the learned features. UFM predicts a convolution kernel and channel-wise modulation cofficients conitioned on the uncertainty weighted representation. Moreover, we develop an uncertainty-driven self-distillation loss to improve the uncertain representation by transferring the knowledge from the confident one. Extensive experimental results on synthetic datasets and real-world images show that UDN achieves significant quantitative and qualitative improvements, outperforming the state-of-the-arts.

A robust instance segmentation framework for underground sewer defect detection

The inspection of underground sewer defects plays a considerable role in estimating the structural integrity and avoiding various unforeseen functional failures. However, the conventional sewer defect inspection approaches suffer from the blurry and vaporous environment inside the sewer pipes, which significantly lowers the performance. Besides, it is challenging to achieve efficient and accurate condition assessment by the common manual inspection. Therefore, this manuscript introduces an automatic instance segmentation-based defect analysis framework. The main contributions include 1) a novel defect segmentation model called Pipe-SOLO is firstly presented to segment six common types of defects at the instance level by proposing an efficient backbone structure (Res2Net-Mish-BN-101) and designing an enhanced BiFPN (EBiFPN), 2) a GAN-based dehazing model is applied to effectively solve the image blurring problem, and 3) a publicly available sewer defect segmentation dataset. The experimental results show the proposed Pipe-SOLO achieved an improvement of 7.3% compared with the state-of-the-art method in terms of the mean Average Precision (mAP). Therefore, the proposed defect segmentation method is promising to be integrated with real-life applications that require defect localization and estimation.

FPD Net: Feature Pyramid DehazeNet

We propose an end-to-end dehazing model based on deep learning (CNN network) and uses the dehazing model re-proposed by AOD-Net based on the atmospheric scattering model for dehazing. Compare to the previously proposed dehazing network, the dehazing model proposed in this paper make use of the FPN network structure in the field of target detection, and uses five feature maps of different sizes to better obtain features of different proportions and different sub-regions. A large amount of experimental data proves that the dehazing model proposed in this paper is superior to previous dehazing technologies in terms of PSNR, SSIM, and subjective visual quality. In addition, it achieved a good performance in speed by using EfficientNet B0 as a feature extractor. We find that only using high-level semantic features can not effectively obtain all the information in the image. The FPN structure used in this paper can effectively integrate the high-level semantics and the low-level semantics, and can better take into account the global and local features. The five feature maps with different sizes are not simply weighted and fused. In order to keep all their information, we put them all together and get the final features through decode layers. At the same time, we have done a comparative experiment between ResNet with FPN and EfficientNet with BiFPN. It is proved that EfficientNet with BiFPN can obtain image features more efficiently. Therefore, EfficientNet with BiFPN is chosen as our network feature extraction.

Contrastive Haze-Aware Learning for Dynamic Remote Sensing Image Dehazing

Image dehazing methods aim to recover a clear image from its hazy counterpart. While various dehazing methods have been proposed, their performance on real-world remote sensing (RS) images remains unsatisfying. A key reason is that the complex weather and imaging conditions (e.g., large fields of view) cause the haze condition to dramatically change in different images, while most existing methods fail to flexibly adapt their dehazing model to the specific haze condition in each image. To mitigate this problem, we present a contrastive haze-aware learning based dynamic dehazing method which demonstrates two aspects of advantage. On one hand, a contrastive clustering scheme is utilized to learn the image-wise haze representation using a set of real-world hazy images in an unsupervised manner, which enables identifying and discriminating the specific haze condition in each given hazy image. On the other hand, with the learned haze representation, a parameter generator can produce haze-aware parameters to dynamically construct a dehazing model for the given hazy image, which empowers us to adaptively dehaze the image based on its specific haze condition and thus improves the generalization ability. In addition, a new contrastive loss defined based on the learned haze representation is further utilized for model training and leads to better performance. To demonstrate the effectiveness of the proposed method, we evaluate it on two benchmark RS image datasets including various real-world hazy images, and observe obviously superiority over other state-of-the-art competitors.

Singe Image Dehazing With Unsharp Masking and Color Gamut Expansion

Image dehazing is a fundamental problem in computer vision and has hitherto engendered prodigious amounts of studies. Recently, with the well-recognized success of deep learning techniques, this field has been dominated by deep dehazing models. However, deep learning is not always a panacea, especially for the practicalities of image dehazing, because high computational complexity, expensive maintenance costs, and high carbon emission are three noticeable problems. Computational efficiency is, therefore, a decisive factor in real-world circumstances. To cope with this growing demand, we propose a linear time algorithm tailored to three primitive parts: unsharp masking (pre-processing), dehazing, and color gamut expansion (post-processing). The first enhances the sharpness according to the local variance of image intensities. The second removes haze based on the improved color attenuation prior, and the third addresses a residual effect of color gamut reduction. Extensive experimental results demonstrated that the proposed method performed comparatively with popular benchmarks, notably deep dehazing models. With such a comparative performance, the proposed method is still fast and efficient, favoring real-world computer vision systems.