Deep Retinex Network for Single Image Dehazing

ODD-Net: a hybrid deep learning architecture for image dehazing

Haze can significantly reduce visibility and contrast of images captured outdoors, necessitating the enhancement of images. This degradation in image quality can adversely affect various applications, including autonomous driving, object detection, and surveillance, where poor visibility may lead to navigation errors and obscure crucial details. Existing dehazing techniques face several challenges: spatial methods tend to be computationally heavy, transform methods often fall short in quality, hybrid methods can be intricate and demanding, and deep learning methods require extensive datasets and computational power. To overcome these challenges, we present ODD-Net, a hybrid deep learning architecture. Our research introduces a comprehensive data set and an innovative architecture called Atmospheric Light Net (A-Net) to estimate atmospheric light, using dilated convolution, batch normalisation, and ReLU activation functions. Furthermore, we develop T-Net to measure information transmission from objects to the camera, using multiscale convolutions and nonlinear regression to create a transmission map. The integrated architecture combines pre-trained A-Net and T-Net models within the atmospheric scattering model. Comparative analysis shows that ODD-Net provides superior dehazing quality, especially in transmission map estimation and depth measurement, surpassing state-of-the-art methods such as DCP, GMAN, DehazeNet, and LCA. Our quantitative analysis reveals that ODD-Net achieves the highest performance in terms the quality metrics compared. The proposed method demonstrates notable quantitative and qualitative improvements over existing techniques, setting a new standard in image dehazing.

TUSR-Net: Triple Unfolding Single Image Dehazing With Self-Regularization and Dual Feature to Pixel Attention

Single image dehazing is a challenging and ill-posed problem due to severe information degeneration of images captured in hazy conditions. Remarkable progresses have been achieved by deep-learning based image dehazing methods, where residual learning is commonly used to separate the hazy image into clear and haze components. However, the nature of low similarity between haze and clear components is commonly neglected, while the lack of constraint of contrastive peculiarity between the two components always restricts the performance of these approaches. To deal with these problems, we propose an end-to-end self-regularized network (TUSR-Net) which exploits the contrastive peculiarity of different components of the hazy image, i.e, self-regularization (SR). In specific, the hazy image is separated into clear and hazy components and constraint between different image components, i.e., self-regularization, is leveraged to pull the recovered clear image closer to groundtruth, which largely promotes the performance of image dehazing. Meanwhile, an effective triple unfolding framework combined with dual feature to pixel attention is proposed to intensify and fuse the intermediate information in feature, channel and pixel levels, respectively, thus features with better representational ability can be obtained. Our TUSR-Net achieves better trade-off between performance and parameter size with weight-sharing strategy and is much more flexible. Experiments on various benchmarking datasets demonstrate the superiority of our TUSR-Net over state-of-the-art single image dehazing methods.

Using Whale Optimization Algorithm and Haze Level Information in a Model-Based Image Dehazing Algorithm

Single image dehazing has been a challenge in the field of image restoration and computer vision. Many model-based and non-model-based dehazing methods have been reported. This study focuses on a model-based algorithm. A popular model-based method is dark channel prior (DCP) which has attracted a lot of attention because of its simplicity and effectiveness. In DCP-based methods, the model parameters should be appropriately estimated for better performance. Previously, we found that appropriate scaling factors of model parameters helped dehazing performance and proposed an improved DCP (IDCP) method that uses heuristic scaling factors for the model parameters (atmospheric light and initial transmittance). With the IDCP, this paper presents an approach to find optimal scaling factors using the whale optimization algorithm (WOA) and haze level information. The WOA uses ground truth images as a reference in a fitness function to search the optimal scaling factors in the IDCP. The IDCP with the WOA was termed IDCP/WOA. It was observed that the performance of IDCP/WOA was significantly affected by hazy ground truth images. Thus, according to the haze level information, a hazy image discriminator was developed to exclude hazy ground truth images from the dataset used in the IDCP/WOA. To avoid using ground truth images in the application stage, hazy image clustering was presented to group hazy images and their corresponding optimal scaling factors obtained by the IDCP/WOA. Then, the average scaling factors for each haze level were found. The resulting dehazing algorithm was called optimized IDCP (OIDCP). Three datasets commonly used in the image dehazing field, the RESIDE, O-HAZE, and KeDeMa datasets, were used to justify the proposed OIDCP. Then a comparison was made between the OIDCP and five recent haze removal methods. On the RESIDE dataset, the OIDCP achieved a PSNR of 26.23 dB, which was better than IDCP by 0.81 dB, DCP by 8.03 dB, RRO by 5.28, AOD by 5.6 dB, and GCAN by 1.27 dB. On the O-HAZE dataset, the OIDCP had a PSNR of 19.53 dB, which was better than IDCP by 0.06 dB, DCP by 4.39 dB, RRO by 0.97 dB, AOD by 1.41 dB, and GCAN by 0.34 dB. On the KeDeMa dataset, the OIDCP obtained the best overall performance and gave dehazed images with stable visual quality. This suggests that the results of this study may benefit model-based dehazing algorithms.

Self-Parameter Distillation Dehazing

In this paper, we propose a novel dehazing method based on self-distillation. In contrast to conventional knowledge distillation approaches that transfer large models (teacher networks) to small models (student networks), we introduce a single knowledge distillation network that transfers network parameters to itself for dehazing. In the early stages, the proposed network transfers scene content (identity) information to the next stage of itself using haze-free data. However, in the later stages, the network transfers haze information to itself using haze data, enabling the accurate dehazing of input images using scene information from the early stages. In a single network, parameters are seamlessly updated from extracting global scene features to dehazing the scene. During the training, forward propagation acts as a teacher network, whereas backward propagation acts as a student network. The experimental results demonstrate that the proposed method considerably outperforms other state-of-the-art dehazing methods.

A Novel Effective Vehicle Detection Method Based on Swin Transformer in Hazy Scenes

Under bad weather, the ability of intelligent vehicles to perceive the environment accurately is an important research content in many practical applications such as smart cities and unmanned driving. In order to improve vehicle environment perception technology in real hazy scenes, we propose an effective detection algorithm based on Swin Transformer for hazy vehicle detection. This algorithm includes two aspects. First of all, for the aspect of the difficulty in extracting haze features with poor visibility, a dehazing network is designed to obtain high-quality haze-free output through encoding and decoding methods using Swin Transformer blocks. In addition, for the aspect of the difficulty of vehicle detection in hazy images, a new end-to-end vehicle detection model in hazy days is constructed by fusing the dehazing module and the Swin Transformer detection module. In the training stage, the self-made dataset Haze-Car is used, and the haze detection model parameters are initialized by using the dehazing model and Swin-T through transfer learning. Finally, the final haze detection model is obtained by fine tuning. Through the joint learning of dehazing and object detection and comparative experiments on the self-made real hazy image dataset, it can be seen that the detection performance of the model in real-world scenes is improved by 12.5%.

Non-Homogeneous Haze Removal via Artificial Scene Prior and Bidimensional Graph Reasoning

Due to the lack of natural scene and haze prior information, it is greatly challenging to completely remove the haze from a single image without distorting its visual content. Fortunately, the real-world haze usually presents non-homogeneous distribution, which provides us with many valuable clues in partial well-preserved regions. In this paper, we propose a Non-Homogeneous Haze Removal Network (NHRN) via artificial scene prior and bidimensional graph reasoning. Firstly, we employ the gamma correction iteratively to simulate artificial multiple shots under different exposure conditions, whose haze degrees are different and enrich the underlying scene prior. Secondly, beyond utilizing the local neighboring relationship, we build a bidimensional graph reasoning module to conduct non-local filtering in the spatial and channel dimensions of feature maps, which models their long-range dependency and propagates the natural scene prior between the well-preserved nodes and the nodes contaminated by haze. To the best of our knowledge, this is the first exploration to remove non-homogeneous haze via the graph reasoning based framework. We evaluate our method on different benchmark datasets. The results demonstrate that our method achieves superior performance over many state-of-the-art algorithms for both the single image dehazing and hazy image understanding tasks. The source code of the proposed NHRN is available on https://github.com/whrws/NHRNet.