This paper introduces a lightweight remote sensing image dehazing network called multidimensional weight regulation network(MDWR-Net), which addresses the high computational cost of existing methods. Previous works, o...This paper introduces a lightweight remote sensing image dehazing network called multidimensional weight regulation network(MDWR-Net), which addresses the high computational cost of existing methods. Previous works, often based on the encoder-decoder structure and utilizing multiple upsampling and downsampling layers, are computationally expensive. To improve efficiency, the paper proposes two modules: the efficient spatial resolution recovery module(ESRR) for upsampling and the efficient depth information augmentation module(EDIA) for downsampling.These modules not only reduce model complexity but also enhance performance. Additionally, the partial feature weight learning module(PFWL) is introduced to reduce the computational burden by applying weight learning across partial dimensions, rather than using full-channel convolution.To overcome the limitations of convolutional neural networks(CNN)-based networks, the haze distribution index transformer(HDIT) is integrated into the decoder. We also propose the physicalbased non-adjacent feature fusion module(PNFF), which leverages the atmospheric scattering model to improve generalization of our MDWR-Net. The MDWR-Net achieves superior dehazing performance with a computational cost of just 2.98×10^(9) multiply-accumulate operations(MACs),which is less than one-tenth of previous methods. Experimental results validate its effectiveness in balancing performance and computational efficiency.展开更多
Although the image dehazing problem has received considerable attention over recent years,the existing models often prioritise performance at the expense of complexity,making them unsuitable for real-world application...Although the image dehazing problem has received considerable attention over recent years,the existing models often prioritise performance at the expense of complexity,making them unsuitable for real-world applications,which require algorithms to be deployed on resource constrained-devices.To address this challenge,we propose WaveLiteDehaze-Network(WLD-Net),an end-to-end dehazing model that delivers performance comparable to complex models while operating in real time and using significantly fewer parameters.This approach capitalises on the insight that haze predominantly affects low-frequency infor-mation.By exclusively processing the image in the frequency domain using discrete wavelet transform(DWT),we segregate the image into high and low frequencies and process them separately.This allows us to preserve high-frequency details and recover low-frequency components affected by haze,distinguishing our method from existing approaches that use spatial domain processing as the backbone,with DWT serving as an auxiliary component.DWT is applied at multiple levels for better in-formation retention while also accelerating computation by downsampling feature maps.Subsequently,a learning-based fusion mechanism reintegrates the processed frequencies to reconstruct the dehazed image.Experiments show that WLD-Net out-performs other low-parameter models on real-world hazy images and rivals much larger models,achieving the highest PSNR and SSIM scores on the O-Haze dataset.Qualitatively,the proposed method demonstrates its effectiveness in handling a diverse range of haze types,delivering visually pleasing results and robust performance,while also generalising well across different scenarios.With only 0.385 million parameters(more than 100 times smaller than comparable dehazing methods),WLD-Net processes 1024×1024 images in just 0.045 s,highlighting its applicability across various real-world scenarios.The code is available at https://github.com/AliMurtaza29/WLD-Net.展开更多
针对当前去雾算法效率不高、细节恢复较差等问题,提出一种改进多尺度AOD-Net(all in one dehazing network)的去雾算法。通过增加注意力机制、调整网络结构和改变损失函数这3方面的改进,增强网络的特征提取和恢复能力。模型的第1层增加...针对当前去雾算法效率不高、细节恢复较差等问题,提出一种改进多尺度AOD-Net(all in one dehazing network)的去雾算法。通过增加注意力机制、调整网络结构和改变损失函数这3方面的改进,增强网络的特征提取和恢复能力。模型的第1层增加空间金字塔注意力(spatial pyramid attention,SPA)机制,使网络在特征提取过程中避免冗余信息。将网络改成拉普拉斯金字塔型结构,使模型能够提取不同尺度的特征,保留特征图的高频信息。使用多尺度结构相似性(multi-scale structural similarity,MS-SSIM)+L1损失函数替换原有的损失函数,提高模型保留结构的能力。实验结果表明,本方法去雾效果更好,细节更丰富。在定性可视化评价方面,去雾图像效果优于原网络。在定量评估层面,与原网络相比PSNR值提升了2.55 dB,SSIM值提升了0.04,IE熵值增加了0.18,这些数值指标充分验证了本算法的出色去雾效果和稳定性。展开更多
The degradation of optical remote sensing images due to atmospheric haze poses a significant obstacle,profoundly impeding their effective utilization across various domains.Dehazing methodologies have emerged as pivot...The degradation of optical remote sensing images due to atmospheric haze poses a significant obstacle,profoundly impeding their effective utilization across various domains.Dehazing methodologies have emerged as pivotal components of image preprocessing,fostering an improvement in the quality of remote sensing imagery.This enhancement renders remote sensing data more indispensable,thereby enhancing the accuracy of target iden-tification.Conventional defogging techniques based on simplistic atmospheric degradation models have proven inadequate for mitigating non-uniform haze within remotely sensed images.In response to this challenge,a novel UNet Residual Attention Network(URA-Net)is proposed.This paradigmatic approach materializes as an end-to-end convolutional neural network distinguished by its utilization of multi-scale dense feature fusion clusters and gated jump connections.The essence of our methodology lies in local feature fusion within dense residual clusters,enabling the extraction of pertinent features from both preceding and current local data,depending on contextual demands.The intelligently orchestrated gated structures facilitate the propagation of these features to the decoder,resulting in superior outcomes in haze removal.Empirical validation through a plethora of experiments substantiates the efficacy of URA-Net,demonstrating its superior performance compared to existing methods when applied to established datasets for remote sensing image defogging.On the RICE-1 dataset,URA-Net achieves a Peak Signal-to-Noise Ratio(PSNR)of 29.07 dB,surpassing the Dark Channel Prior(DCP)by 11.17 dB,the All-in-One Network for Dehazing(AOD)by 7.82 dB,the Optimal Transmission Map and Adaptive Atmospheric Light For Dehazing(OTM-AAL)by 5.37 dB,the Unsupervised Single Image Dehazing(USID)by 8.0 dB,and the Superpixel-based Remote Sensing Image Dehazing(SRD)by 8.5 dB.Particularly noteworthy,on the SateHaze1k dataset,URA-Net attains preeminence in overall performance,yielding defogged images characterized by consistent visual quality.This underscores the contribution of the research to the advancement of remote sensing technology,providing a robust and efficient solution for alleviating the adverse effects of haze on image quality.展开更多
In order to solve the problems of color bias and visual deviation caused by inaccurate estimation of transmittance and atmospheric light in image defogging,a new algorithm based on multi-scale morphological reconstruc...In order to solve the problems of color bias and visual deviation caused by inaccurate estimation of transmittance and atmospheric light in image defogging,a new algorithm based on multi-scale morphological reconstruction with adaptive transmittance and atmospheric light correction was proposed.Firstly,the algorithm used the open operation under morphological reconstruction to replace the minimum filter operation in the dark channel,and used the morphological edge to set the scale of the open operation structure elements,and constructed a multi-scale open operation fusion dark channel.After morphological noise reduction,the exact initial transmittance was obtained.According to the relationship between brightness and saturation difference and transmittance,an adaptive transmittance correction model was fitted with Gaussian function to correct the initial transmittance of the sky fog map.Then the local atmospheric light was improved according to the image brightness information and morphology closure operation.Finally,the proposed algorithm was combined with the atmospheric scattering model to obtain an accurate fog free image.The experimental results showed that the proposed algorithm was suitable for fog image restoration under various scenes,the restoration effect was good,and the brightness was suitable.展开更多
Due to the presence of turbid media, such as microdust and water vapor in the environment, outdoor pictures taken under hazy weather circumstances are typically degraded. To enhance the quality of such images, this wo...Due to the presence of turbid media, such as microdust and water vapor in the environment, outdoor pictures taken under hazy weather circumstances are typically degraded. To enhance the quality of such images, this work proposes a new hybrid λ2-λ0 penalty model for image dehazing. This model performs a weighted fusion of two distinct transmission maps, generated by imposing λ2 and λ0 norm penalties on the approximate regression coefficients of the transmission map. This approach effectively balances the sparsity and smoothness associated with the λ0 and λ2 norms, thereby optimizing the transmittance map. Specifically, when the λ2 norm is penalized in the model, an updated guided image is obtained after implementing λ0 penalty. The resulting optimization problem is effectively solved using the least square method and the alternating direction algorithm. The dehazing framework combines the advantages of λ2 and λ0 norms, enhancing sparse and smoothness, resulting in higher quality images with clearer details and preserved edges.展开更多
经典AOD-Net(All in One Dehazing Network)去雾后的图像存在细节清晰度不足、明暗反差过大和画面昏暗等问题。为了解决这些图像去雾问题,提出一种在AOD-Net基础上改进的多尺度算法。改进的网络结构采用深度可分离卷积替换传统卷积方式...经典AOD-Net(All in One Dehazing Network)去雾后的图像存在细节清晰度不足、明暗反差过大和画面昏暗等问题。为了解决这些图像去雾问题,提出一种在AOD-Net基础上改进的多尺度算法。改进的网络结构采用深度可分离卷积替换传统卷积方式,减少了冗余参数量,加快了计算速度并有效地减少了模型的内存占用量,从而提高了算法去雾效率;同时采用多尺度结构在不同尺度上对雾图进行分析和处理,更好地捕捉图像的细节信息,提升了网络对图像细节的处理能力,解决了原算法去雾时存在的细节模糊问题;最后在网络结构中加入金字塔池化模块,用于整合图像不同区域的上下文信息,扩展了网络的感知范围,从而提高网络模型获取有雾图像全局信息的能力,进而改善图像色调失真、细节丢失等问题。此外,引入一个低照度增强模块,通过明确预测噪声实现去噪的目标,从而恢复曝光不足的图像。在低光去雾图像中,峰值信噪比(PSNR)和结构相似性(SSIM)指标均有显著提升,处理后的图片具有更高的整体自然度。实验结果表明:与经典AOD-Net去雾的结果相比,改进算法能够更好地恢复图像的细节和结构,使得去雾后的图像更自然,饱和度和对比度也更加平衡;在RESIDE的SOTS数据集中的室外和室内场景,相较于经典AOD-Net,改进算法的PSNR分别提升了4.5593 dB和4.0656 dB,SSIM分别提升了0.0476和0.0874。展开更多
Images captured in hazy or foggy weather conditions can be seriously degraded by scattering of atmospheric particles,which reduces the contrast,changes the color,and makes the object features difficult to identify by ...Images captured in hazy or foggy weather conditions can be seriously degraded by scattering of atmospheric particles,which reduces the contrast,changes the color,and makes the object features difficult to identify by human vision and by some outdoor computer vision systems.Therefore image dehazing is an important issue and has been widely researched in the field of computer vision.The role of image dehazing is to remove the influence of weather factors in order to improve the visual effects of the image and provide benefit to post-processing.This paper reviews the main techniques of image dehazing that have been developed over the past decade.Firstly,we innovatively divide a number of approaches into three categories:image enhancement based methods,image fusion based methods and image restoration based methods.All methods are analyzed and corresponding sub-categories are introduced according to principles and characteristics.Various quality evaluation methods are then described,sorted and discussed in detail.Finally,research progress is summarized and future research directions are suggested.展开更多
We present two haze removal algorithms for single image based on haziness analysis.One algorithm regards haze as the veil layer,and the other takes haze as the transmission.The former uses the illumination component i...We present two haze removal algorithms for single image based on haziness analysis.One algorithm regards haze as the veil layer,and the other takes haze as the transmission.The former uses the illumination component image obtained by retinex algorithm and the depth information of the original image to remove the veil layer.The latter employs guided filter to obtain the refined haze transmission and separates it from the original image.The main advantages of the proposed methods are that no user interaction is needed and the computing speed is relatively fast.A comparative study and quantitative evaluation with some main existing algorithms demonstrate that similar even better quality results can be obtained by the proposed methods.On the top of haze removal,several applications of the haze transmission including image refocusing,haze simulation,relighting and 2-dimensional(2D)to 3-dimensional(3D) stereoscopic conversion are also implemented.展开更多
In this paper,we propose a neoteric and high-efficiency single image dehazing algorithm via contrast enhancement which is called STRASS(Spatio-Temporal Retinex-Inspired by an Averaging of Stochastic Samples)dehazing,i...In this paper,we propose a neoteric and high-efficiency single image dehazing algorithm via contrast enhancement which is called STRASS(Spatio-Temporal Retinex-Inspired by an Averaging of Stochastic Samples)dehazing,it is realized by constructing an efficient high-pass filter to process haze images and taking the influence of human vision system into account in image dehazing principles.The novel high-pass filter works by getting each pixel using RSR and computes the average of the samples.Then the low-pass filter resulting from the minimum envelope in STRESS framework has been replaced by the average of the samples.The final dehazed image is yielded after iterations of the high-pass filter.STRASS can be run directly without any machine learning.Extensive experimental results on datasets prove that STRASS surpass the state-of-the-arts.Image dehazing can be applied in the field of printing and packaging,our method is of great significance for image pre-processing before printing.展开更多
Low visibility in foggy days results in less contrasted and blurred images with color distortion which adversely affects and leads to the sub-optimal performances in image and video monitoring systems. The causes of f...Low visibility in foggy days results in less contrasted and blurred images with color distortion which adversely affects and leads to the sub-optimal performances in image and video monitoring systems. The causes of foggy image degradation were explained in detail and the approaches of image enhancement and image restoration for defogging were introduced. The study proposed an enhanced and advanced form of the improved Retinex theory-based dehazing algorithm. The proposed algorithm achieved novel in the manner in which the dark channel prior was efficiently combined with the dark-channel prior into a single dehazing framework. The proposed approach performed the first stage in dehazing within the dark channel domain through implementation with an adaptive filter. This novel approach allowed for the dark channel features to be efficiently refined and boosted, a scheme, which according to the obtained results, significantly improved dehazing results in later stages. Experimental results showed that this approach did little to trade-off dehazing speed for efficiency. This makes the proposed algorithm a strong candidate for real-time systems due to its capability to realize efficient dehazing at considerably rapid speeds. Finally, experimental results were provided to validate the superior performance and efficiency of the proposed dehazing algorithm.展开更多
This paper presents a trainable Generative Adversarial Network(GAN)-based end-to-end system for image dehazing,which is named the DehazeGAN.DehazeGAN can be used for edge computing-based applications,such as roadside ...This paper presents a trainable Generative Adversarial Network(GAN)-based end-to-end system for image dehazing,which is named the DehazeGAN.DehazeGAN can be used for edge computing-based applications,such as roadside monitoring.It adopts two networks:one is generator(G),and the other is discriminator(D).The G adopts the U-Net architecture,whose layers are particularly designed to incorporate the atmospheric scattering model of image dehazing.By using a reformulated atmospheric scattering model,the weights of the generator network are initialized by the coarse transmission map,and the biases are adaptively adjusted by using the previous round's trained weights.Since the details may be blurry after the fog is removed,the contrast loss is added to enhance the visibility actively.Aside from the typical GAN adversarial loss,the pixel-wise Mean Square Error(MSE)loss,the contrast loss and the dark channel loss are introduced into the generator loss function.Extensive experiments on benchmark images,the results of which are compared with those of several state-of-the-art methods,demonstrate that the proposed DehazeGAN performs better and is more effective.展开更多
As one of the most simple and effective single image dehazing methods, the dark channel prior(DCP) algorithm has been widely applied. However, the algorithm does not work for pixels similar to airlight(e.g., snowy gro...As one of the most simple and effective single image dehazing methods, the dark channel prior(DCP) algorithm has been widely applied. However, the algorithm does not work for pixels similar to airlight(e.g., snowy ground or a white wall), resulting in underestimation of the transmittance of some local scenes. To address that problem, we propose an image dehazing method by incorporating Markov random field(MRF) with the DCP. The DCP explicitly represents the input image observation in the MRF model obtained by the transmittance map. The key idea is that the sparsely distributed wrongly estimated transmittance can be corrected by properly characterizing the spatial dependencies between the neighboring pixels of the transmittances that are well estimated and those that are wrongly estimated. To that purpose, the energy function of the MRF model is designed. The estimation of the initial transmittance map is pixel-based using the DCP, and the segmentation on the transmittance map is employed to separate the foreground and background, thereby avoiding the block effect and artifacts at the depth discontinuity. Given the limited number of labels obtained by clustering, the smoothing term in the MRF model can properly smooth the transmittance map without an extra refinement filter. Experimental results obtained by using terrestrial and underwater images are given.展开更多
The captured outdoor images and videos may appear blurred due to haze,fog,and bad weather conditions.Water droplets or dust particles in the atmosphere cause the light to scatter,resulting in very limited scene discer...The captured outdoor images and videos may appear blurred due to haze,fog,and bad weather conditions.Water droplets or dust particles in the atmosphere cause the light to scatter,resulting in very limited scene discernibility and deterioration in the quality of the image captured.Currently,image dehazing has gainedmuch popularity because of its usability in a wide variety of applications.Various algorithms have been proposed to solve this ill-posed problem.These algorithms provide quite promising results in some cases,but they include undesirable artifacts and noise in haze patches in adverse cases.Some of these techniques take unrealistic processing time for high image resolution.In this paper,to achieve real-time halo-free dehazing,fast and effective single image dehazing we propose a simple but effective image restoration technique using multiple patches.It will improve the shortcomings of DCP and improve its speed and efficiency for high-resolution images.A coarse transmissionmap is estimated by using the minimumof different size patches.Then a cascaded fast guided filter is used to refine the transmission map.We introduce an efficient scaling technique for transmission map estimation,which gives an advantage of very low-performance degradation for a highresolution image.For performance evaluation,quantitative,qualitative and computational time comparisons have been performed,which provide quiet faithful results in speed,quality,and reliability of handling bright surfaces.展开更多
In the field of computer and machine vision, haze and fog lead to image degradation through various degradation mechanisms including but not limited to contrast attenuation, blurring and pixel distortions. This limits...In the field of computer and machine vision, haze and fog lead to image degradation through various degradation mechanisms including but not limited to contrast attenuation, blurring and pixel distortions. This limits the efficiency of machine vision systems such as video surveillance, target tracking and recognition. Various single image dark channel dehazing algorithms have aimed to tackle the problem of image hazing in a fast and efficient manner. Such algorithms rely upon the dark channel prior theory towards the estimation of the atmospheric light which offers itself as a crucial parameter towards dehazing. This paper studies the state-of-the-art in this area and puts forwards their strengths and weaknesses. Through experiments the efficiencies and shortcomings of these algorithms are shared. This information is essential for researchers and developers in providing a reference for the development of applications and future of the research field.展开更多
文摘This paper introduces a lightweight remote sensing image dehazing network called multidimensional weight regulation network(MDWR-Net), which addresses the high computational cost of existing methods. Previous works, often based on the encoder-decoder structure and utilizing multiple upsampling and downsampling layers, are computationally expensive. To improve efficiency, the paper proposes two modules: the efficient spatial resolution recovery module(ESRR) for upsampling and the efficient depth information augmentation module(EDIA) for downsampling.These modules not only reduce model complexity but also enhance performance. Additionally, the partial feature weight learning module(PFWL) is introduced to reduce the computational burden by applying weight learning across partial dimensions, rather than using full-channel convolution.To overcome the limitations of convolutional neural networks(CNN)-based networks, the haze distribution index transformer(HDIT) is integrated into the decoder. We also propose the physicalbased non-adjacent feature fusion module(PNFF), which leverages the atmospheric scattering model to improve generalization of our MDWR-Net. The MDWR-Net achieves superior dehazing performance with a computational cost of just 2.98×10^(9) multiply-accumulate operations(MACs),which is less than one-tenth of previous methods. Experimental results validate its effectiveness in balancing performance and computational efficiency.
基金Japan International Cooperation Agency(JICA)via Malaysia-Japan Linkage Research Grant 2024.
文摘Although the image dehazing problem has received considerable attention over recent years,the existing models often prioritise performance at the expense of complexity,making them unsuitable for real-world applications,which require algorithms to be deployed on resource constrained-devices.To address this challenge,we propose WaveLiteDehaze-Network(WLD-Net),an end-to-end dehazing model that delivers performance comparable to complex models while operating in real time and using significantly fewer parameters.This approach capitalises on the insight that haze predominantly affects low-frequency infor-mation.By exclusively processing the image in the frequency domain using discrete wavelet transform(DWT),we segregate the image into high and low frequencies and process them separately.This allows us to preserve high-frequency details and recover low-frequency components affected by haze,distinguishing our method from existing approaches that use spatial domain processing as the backbone,with DWT serving as an auxiliary component.DWT is applied at multiple levels for better in-formation retention while also accelerating computation by downsampling feature maps.Subsequently,a learning-based fusion mechanism reintegrates the processed frequencies to reconstruct the dehazed image.Experiments show that WLD-Net out-performs other low-parameter models on real-world hazy images and rivals much larger models,achieving the highest PSNR and SSIM scores on the O-Haze dataset.Qualitatively,the proposed method demonstrates its effectiveness in handling a diverse range of haze types,delivering visually pleasing results and robust performance,while also generalising well across different scenarios.With only 0.385 million parameters(more than 100 times smaller than comparable dehazing methods),WLD-Net processes 1024×1024 images in just 0.045 s,highlighting its applicability across various real-world scenarios.The code is available at https://github.com/AliMurtaza29/WLD-Net.
文摘针对当前去雾算法效率不高、细节恢复较差等问题,提出一种改进多尺度AOD-Net(all in one dehazing network)的去雾算法。通过增加注意力机制、调整网络结构和改变损失函数这3方面的改进,增强网络的特征提取和恢复能力。模型的第1层增加空间金字塔注意力(spatial pyramid attention,SPA)机制,使网络在特征提取过程中避免冗余信息。将网络改成拉普拉斯金字塔型结构,使模型能够提取不同尺度的特征,保留特征图的高频信息。使用多尺度结构相似性(multi-scale structural similarity,MS-SSIM)+L1损失函数替换原有的损失函数,提高模型保留结构的能力。实验结果表明,本方法去雾效果更好,细节更丰富。在定性可视化评价方面,去雾图像效果优于原网络。在定量评估层面,与原网络相比PSNR值提升了2.55 dB,SSIM值提升了0.04,IE熵值增加了0.18,这些数值指标充分验证了本算法的出色去雾效果和稳定性。
基金This project is supported by the National Natural Science Foundation of China(NSFC)(No.61902158).
文摘The degradation of optical remote sensing images due to atmospheric haze poses a significant obstacle,profoundly impeding their effective utilization across various domains.Dehazing methodologies have emerged as pivotal components of image preprocessing,fostering an improvement in the quality of remote sensing imagery.This enhancement renders remote sensing data more indispensable,thereby enhancing the accuracy of target iden-tification.Conventional defogging techniques based on simplistic atmospheric degradation models have proven inadequate for mitigating non-uniform haze within remotely sensed images.In response to this challenge,a novel UNet Residual Attention Network(URA-Net)is proposed.This paradigmatic approach materializes as an end-to-end convolutional neural network distinguished by its utilization of multi-scale dense feature fusion clusters and gated jump connections.The essence of our methodology lies in local feature fusion within dense residual clusters,enabling the extraction of pertinent features from both preceding and current local data,depending on contextual demands.The intelligently orchestrated gated structures facilitate the propagation of these features to the decoder,resulting in superior outcomes in haze removal.Empirical validation through a plethora of experiments substantiates the efficacy of URA-Net,demonstrating its superior performance compared to existing methods when applied to established datasets for remote sensing image defogging.On the RICE-1 dataset,URA-Net achieves a Peak Signal-to-Noise Ratio(PSNR)of 29.07 dB,surpassing the Dark Channel Prior(DCP)by 11.17 dB,the All-in-One Network for Dehazing(AOD)by 7.82 dB,the Optimal Transmission Map and Adaptive Atmospheric Light For Dehazing(OTM-AAL)by 5.37 dB,the Unsupervised Single Image Dehazing(USID)by 8.0 dB,and the Superpixel-based Remote Sensing Image Dehazing(SRD)by 8.5 dB.Particularly noteworthy,on the SateHaze1k dataset,URA-Net attains preeminence in overall performance,yielding defogged images characterized by consistent visual quality.This underscores the contribution of the research to the advancement of remote sensing technology,providing a robust and efficient solution for alleviating the adverse effects of haze on image quality.
基金supported by National Natural Science Foundation of China(No.61561030)College Industry Support Plan Project of Gansu Provincial Department of Education(No.2021CYZC-04)Educational Reform Fund of Lanzhou Jiaotong University(No.JG201928)。
文摘In order to solve the problems of color bias and visual deviation caused by inaccurate estimation of transmittance and atmospheric light in image defogging,a new algorithm based on multi-scale morphological reconstruction with adaptive transmittance and atmospheric light correction was proposed.Firstly,the algorithm used the open operation under morphological reconstruction to replace the minimum filter operation in the dark channel,and used the morphological edge to set the scale of the open operation structure elements,and constructed a multi-scale open operation fusion dark channel.After morphological noise reduction,the exact initial transmittance was obtained.According to the relationship between brightness and saturation difference and transmittance,an adaptive transmittance correction model was fitted with Gaussian function to correct the initial transmittance of the sky fog map.Then the local atmospheric light was improved according to the image brightness information and morphology closure operation.Finally,the proposed algorithm was combined with the atmospheric scattering model to obtain an accurate fog free image.The experimental results showed that the proposed algorithm was suitable for fog image restoration under various scenes,the restoration effect was good,and the brightness was suitable.
文摘Due to the presence of turbid media, such as microdust and water vapor in the environment, outdoor pictures taken under hazy weather circumstances are typically degraded. To enhance the quality of such images, this work proposes a new hybrid λ2-λ0 penalty model for image dehazing. This model performs a weighted fusion of two distinct transmission maps, generated by imposing λ2 and λ0 norm penalties on the approximate regression coefficients of the transmission map. This approach effectively balances the sparsity and smoothness associated with the λ0 and λ2 norms, thereby optimizing the transmittance map. Specifically, when the λ2 norm is penalized in the model, an updated guided image is obtained after implementing λ0 penalty. The resulting optimization problem is effectively solved using the least square method and the alternating direction algorithm. The dehazing framework combines the advantages of λ2 and λ0 norms, enhancing sparse and smoothness, resulting in higher quality images with clearer details and preserved edges.
文摘经典AOD-Net(All in One Dehazing Network)去雾后的图像存在细节清晰度不足、明暗反差过大和画面昏暗等问题。为了解决这些图像去雾问题,提出一种在AOD-Net基础上改进的多尺度算法。改进的网络结构采用深度可分离卷积替换传统卷积方式,减少了冗余参数量,加快了计算速度并有效地减少了模型的内存占用量,从而提高了算法去雾效率;同时采用多尺度结构在不同尺度上对雾图进行分析和处理,更好地捕捉图像的细节信息,提升了网络对图像细节的处理能力,解决了原算法去雾时存在的细节模糊问题;最后在网络结构中加入金字塔池化模块,用于整合图像不同区域的上下文信息,扩展了网络的感知范围,从而提高网络模型获取有雾图像全局信息的能力,进而改善图像色调失真、细节丢失等问题。此外,引入一个低照度增强模块,通过明确预测噪声实现去噪的目标,从而恢复曝光不足的图像。在低光去雾图像中,峰值信噪比(PSNR)和结构相似性(SSIM)指标均有显著提升,处理后的图片具有更高的整体自然度。实验结果表明:与经典AOD-Net去雾的结果相比,改进算法能够更好地恢复图像的细节和结构,使得去雾后的图像更自然,饱和度和对比度也更加平衡;在RESIDE的SOTS数据集中的室外和室内场景,相较于经典AOD-Net,改进算法的PSNR分别提升了4.5593 dB和4.0656 dB,SSIM分别提升了0.0476和0.0874。
基金supported by the National Natural Science Foundation of China(61403283)Shandong Provincial Natural Science Foundation(ZR2013FQ036.ZR2015PE025)+2 种基金the Spark Program of China(2013GA740053)the Spark Program of Shandong Province(2013XH06034)the Technology Development Plan of Weifang City(201301015)
文摘Images captured in hazy or foggy weather conditions can be seriously degraded by scattering of atmospheric particles,which reduces the contrast,changes the color,and makes the object features difficult to identify by human vision and by some outdoor computer vision systems.Therefore image dehazing is an important issue and has been widely researched in the field of computer vision.The role of image dehazing is to remove the influence of weather factors in order to improve the visual effects of the image and provide benefit to post-processing.This paper reviews the main techniques of image dehazing that have been developed over the past decade.Firstly,we innovatively divide a number of approaches into three categories:image enhancement based methods,image fusion based methods and image restoration based methods.All methods are analyzed and corresponding sub-categories are introduced according to principles and characteristics.Various quality evaluation methods are then described,sorted and discussed in detail.Finally,research progress is summarized and future research directions are suggested.
基金supported by National Natural Science Foundation of China(Nos.91220301,61175064 and 61273314)Postdoctoral Science Foundation of Central South University(No.126648)New Teacher Fund for School of Information Science and Engineering,Central South University(No.2012170301)
文摘We present two haze removal algorithms for single image based on haziness analysis.One algorithm regards haze as the veil layer,and the other takes haze as the transmission.The former uses the illumination component image obtained by retinex algorithm and the depth information of the original image to remove the veil layer.The latter employs guided filter to obtain the refined haze transmission and separates it from the original image.The main advantages of the proposed methods are that no user interaction is needed and the computing speed is relatively fast.A comparative study and quantitative evaluation with some main existing algorithms demonstrate that similar even better quality results can be obtained by the proposed methods.On the top of haze removal,several applications of the haze transmission including image refocusing,haze simulation,relighting and 2-dimensional(2D)to 3-dimensional(3D) stereoscopic conversion are also implemented.
基金This work was supported in part by National Natural Science Foundation of China under Grant 62076199in part by the Open Research Fund of Beijing Key Laboratory of Big Data Technology for Food Safety under Grant BTBD-2020KF08+2 种基金Beijing Technology and Business University,in part by the China Postdoctoral Science Foundation under Grant 2019M653784in part by Key Laboratory of Spectral Imaging Technology of Chinese Academy of Sciences under Grant LSIT201801Din part by the Key R&D Project of Shaan’xi Province under Grant 2021GY-027。
文摘In this paper,we propose a neoteric and high-efficiency single image dehazing algorithm via contrast enhancement which is called STRASS(Spatio-Temporal Retinex-Inspired by an Averaging of Stochastic Samples)dehazing,it is realized by constructing an efficient high-pass filter to process haze images and taking the influence of human vision system into account in image dehazing principles.The novel high-pass filter works by getting each pixel using RSR and computes the average of the samples.Then the low-pass filter resulting from the minimum envelope in STRESS framework has been replaced by the average of the samples.The final dehazed image is yielded after iterations of the high-pass filter.STRASS can be run directly without any machine learning.Extensive experimental results on datasets prove that STRASS surpass the state-of-the-arts.Image dehazing can be applied in the field of printing and packaging,our method is of great significance for image pre-processing before printing.
文摘Low visibility in foggy days results in less contrasted and blurred images with color distortion which adversely affects and leads to the sub-optimal performances in image and video monitoring systems. The causes of foggy image degradation were explained in detail and the approaches of image enhancement and image restoration for defogging were introduced. The study proposed an enhanced and advanced form of the improved Retinex theory-based dehazing algorithm. The proposed algorithm achieved novel in the manner in which the dark channel prior was efficiently combined with the dark-channel prior into a single dehazing framework. The proposed approach performed the first stage in dehazing within the dark channel domain through implementation with an adaptive filter. This novel approach allowed for the dark channel features to be efficiently refined and boosted, a scheme, which according to the obtained results, significantly improved dehazing results in later stages. Experimental results showed that this approach did little to trade-off dehazing speed for efficiency. This makes the proposed algorithm a strong candidate for real-time systems due to its capability to realize efficient dehazing at considerably rapid speeds. Finally, experimental results were provided to validate the superior performance and efficiency of the proposed dehazing algorithm.
基金This research was supported by Basic Science Research Program through the National Research Foundation of Korea(NRF)funded by the Ministry of Education(grant number NRF-2018R1D1A1B07043331).
文摘This paper presents a trainable Generative Adversarial Network(GAN)-based end-to-end system for image dehazing,which is named the DehazeGAN.DehazeGAN can be used for edge computing-based applications,such as roadside monitoring.It adopts two networks:one is generator(G),and the other is discriminator(D).The G adopts the U-Net architecture,whose layers are particularly designed to incorporate the atmospheric scattering model of image dehazing.By using a reformulated atmospheric scattering model,the weights of the generator network are initialized by the coarse transmission map,and the biases are adaptively adjusted by using the previous round's trained weights.Since the details may be blurry after the fog is removed,the contrast loss is added to enhance the visibility actively.Aside from the typical GAN adversarial loss,the pixel-wise Mean Square Error(MSE)loss,the contrast loss and the dark channel loss are introduced into the generator loss function.Extensive experiments on benchmark images,the results of which are compared with those of several state-of-the-art methods,demonstrate that the proposed DehazeGAN performs better and is more effective.
基金supported by the National Natural Science Foundation of China (No.61571407)。
文摘As one of the most simple and effective single image dehazing methods, the dark channel prior(DCP) algorithm has been widely applied. However, the algorithm does not work for pixels similar to airlight(e.g., snowy ground or a white wall), resulting in underestimation of the transmittance of some local scenes. To address that problem, we propose an image dehazing method by incorporating Markov random field(MRF) with the DCP. The DCP explicitly represents the input image observation in the MRF model obtained by the transmittance map. The key idea is that the sparsely distributed wrongly estimated transmittance can be corrected by properly characterizing the spatial dependencies between the neighboring pixels of the transmittances that are well estimated and those that are wrongly estimated. To that purpose, the energy function of the MRF model is designed. The estimation of the initial transmittance map is pixel-based using the DCP, and the segmentation on the transmittance map is employed to separate the foreground and background, thereby avoiding the block effect and artifacts at the depth discontinuity. Given the limited number of labels obtained by clustering, the smoothing term in the MRF model can properly smooth the transmittance map without an extra refinement filter. Experimental results obtained by using terrestrial and underwater images are given.
基金This research was supported by the MSIT(Ministry of Science and ICT),Korea,under the ICAN(ICT Challenge and Advanced Network of HRD)program(IITP-2021-2020-0-01832)supervised by the IITP(Institute of Information&Communications Technology Planning&Evaluation)and the Soonchunhyang University Research Fund.
文摘The captured outdoor images and videos may appear blurred due to haze,fog,and bad weather conditions.Water droplets or dust particles in the atmosphere cause the light to scatter,resulting in very limited scene discernibility and deterioration in the quality of the image captured.Currently,image dehazing has gainedmuch popularity because of its usability in a wide variety of applications.Various algorithms have been proposed to solve this ill-posed problem.These algorithms provide quite promising results in some cases,but they include undesirable artifacts and noise in haze patches in adverse cases.Some of these techniques take unrealistic processing time for high image resolution.In this paper,to achieve real-time halo-free dehazing,fast and effective single image dehazing we propose a simple but effective image restoration technique using multiple patches.It will improve the shortcomings of DCP and improve its speed and efficiency for high-resolution images.A coarse transmissionmap is estimated by using the minimumof different size patches.Then a cascaded fast guided filter is used to refine the transmission map.We introduce an efficient scaling technique for transmission map estimation,which gives an advantage of very low-performance degradation for a highresolution image.For performance evaluation,quantitative,qualitative and computational time comparisons have been performed,which provide quiet faithful results in speed,quality,and reliability of handling bright surfaces.
文摘In the field of computer and machine vision, haze and fog lead to image degradation through various degradation mechanisms including but not limited to contrast attenuation, blurring and pixel distortions. This limits the efficiency of machine vision systems such as video surveillance, target tracking and recognition. Various single image dark channel dehazing algorithms have aimed to tackle the problem of image hazing in a fast and efficient manner. Such algorithms rely upon the dark channel prior theory towards the estimation of the atmospheric light which offers itself as a crucial parameter towards dehazing. This paper studies the state-of-the-art in this area and puts forwards their strengths and weaknesses. Through experiments the efficiencies and shortcomings of these algorithms are shared. This information is essential for researchers and developers in providing a reference for the development of applications and future of the research field.
