A joint cumulative distribution function and gradient fusion based method for dehazing of long shot hazy images

Hazy or foggy weather conditions significantly degrade the visual quality of an image in an outdoor environment. It also changes the color and reduces the contrast of an image. This paper introduces a novel single image dehazing technique to restore a hazy image without considering the physical model of haze formation. In order to find haze-free image, the proposed method does not require the transmission map and its costly refinement process. Since haze effect is dependent on the depth, it severely degrades the visibility of the objects located at a far distance. The objects close to the camera are unaffected. In this paper, we propose a fusion-based haze removal method based on the joint cumulative distribution function (JCDF) that treats faraway haze and nearby haze separately. The output images after the JCDF module, fused in the gradient domain to produce a haze-free image. The proposed method not only significantly enhances visibility but also preserves texture details. The proposed method is experimented and evaluated on a large set of challenging hazy images (large scene depth, night time, dense fog, etc.). Both qualitative and quantitative measures show that the performance of the proposed method is better than the state-of-the-art dehazing techniques.     

A multi-exposure fusion framework for contrast enhancement of hazy images employing dynamic stochastic resonance

Current imaging devices coupled with advanced hardware and software are smart enough to enhance low light images taken in clear weather. But in hazy or foggy environments, the captured images are of degraded quality. To address this issue, image processing algorithms are employed to enhance the degraded images to make useful for extracting meaningful features. In this study, we propose a haze removal algorithm to improve the color and contrast of images captured in hazy environments. The first step involves generation of images with various exposures using the theory of dynamic stochastic resonance. The images are then fused in a multi-scale fusion framework crafting weight maps viz. haze density, chromaticity, and luminance gradient. The fusion process focuses on uniformly enhancing the dark and bright regions of the image. However, it may overemphasize haze affected regions. Therefore, in the second step, the atmospheric scattering equation is referred and its modified version is applied that accomplishes the haze removal task. Quantitative and qualitative analyses demonstrate the effectiveness of the proposed method.     

尺度自适应暗通道先验去雾方法

在雾、霾等天气条件下,大气粒子的散射作用使环境的能见度偏低,视觉系统采集到的图像严重降质。基于暗通道先验的图像复原方法因其去雾效果自然、约束条件少,且易于实现等优点而受到广泛关注。但是,该方法的去雾效果受尺度（暗通道的求解半径）影响很大,对于不同场景的图像,不存在一个普遍适用的最优尺度。针对该问题,文中提出一种尺度自适应方法,根据图像的颜色和边缘特征自适应地调节暗通道的尺度范围,得到像素级的暗通道求解尺度,兼顾大尺度求解色彩失真小和小尺度求解光晕失真小等优点。此外,针对暗通道去雾方法会使天空光估计点落到前景区域的问题,提出了一种改进的天空光估计方法,可使估计点鲁棒地落到与其物理意义相符的背景区域。对多种雾化场景图像的处理结果表明:文中方法适应性强、去雾效果自然,且对比度提升显著。     

From synthetic to natural — single natural image dehazing deep networks using synthetic dataset domain randomization

Image dehazing methods aim to solve the problem of poor visibility in images due to haze. Techniques proposed for image dehazing in literature focus on image priors, haze lines or data driven statistical models. Variations of the classical methods relying on prior model or haze line model use no-reference image quality metrics to prove their dehazing performance. Recently developed deep learning models rely on huge amounts of hazy, haze-free pairs for training, and uses PSNR and SSIM like image reconstruction metrics to show their performance. These methods perform poorly on no-reference image quality assessments and also dehazes poorly at the depths of the image. These methods though can be optimized for memory usage and are faster. This work presents a deep learning model (Feature Fusion Attention Network) trained on a domain randomized synthetic dataset generated in simulation. The proposed model achieves the highest scores on blind image assessments through the gradient rationing technique for a deep learning-based approach by a significant margin. The images were evaluated on full-reference metrics as well and obtained favorable results. This approach also yields one of the highest edge sharpness obtained after dehazing. The training procedure adopted to obtain significant gains on real-world dehazing, without using any real-world data is also detailed in this paper.     

Image blind restoration based on degradation representation network

Most deep learning (DL)-based image restoration methods have exploited excellent performance by learning a non-linear mapping function from low quality images to high quality images. However, two major problems restrict the development of the image restoration methods. First, most existing methods based on fixed degradation suffer from significant performance drop when facing the unknown degradation, because of the huge gap between the fixed degradation and the unknown degradation. Second, the unknown-degradation estimation may lead to restoration task failure due to uncertain estimation errors. To handle the unknown degradation in the real application, we introduce a degradation representation network for single image blind restoration (DRN). Different from the methods of estimating pixel space, we use an encoder network to learn abstract representations for estimating different degradation kernels in the representation space. Furthermore, a degradation perception module with flexible adaptability to different degradation kernels is used to restore more structural details. In our experiments, we compare our DRN with several state-of-the-art methods for two image restoration tasks, including image super-resolution (SR) and image denoising. Quantitative results show that our degradation representation network is accurate and efficient for single image restoration.     

基于霾层学习的单幅图像去雾算法

针对传统去雾算法出现色彩失真、去雾不完全、出现光晕等现象,本文提出了一种基于霾层学习的卷积神经网络的单幅图像去雾算法.首先,依据大气散射物理模型进行理论推导,本文设计了一种能够直接学习和估计有雾图像和霾层图像之间的映射关系的网络模型.采用有雾图像作为输入,并输出有雾图像与无雾图像之间的残差图像,随后直接从有雾图像中去除此霾层图像,即可恢复出无雾图像.残差学习的引入,使得网络来直接估计初始霾层,利用相对大的学习率,减少计算量,加快收敛过程.再利用引导滤波进行细化,使得恢复出的无雾图像更接近真实场景.本文对不同雾浓度的有雾图片的去雾效果进行测试,并与当前主流深度学习去雾算法及其他经典算法进行对比.实验结果显示,本文设计的卷积神经网络模型在图像去雾的应用,不论在主观效果还是客观指标上,都有优势.     

图像雾霾等级评价及去雾技术研究进展

图像去雾技术是对有雾图像进行清晰化处理的一门技术,该技术的任务是去除环境因素对图像质量的影响,从而增强图像的视见度。文章首先对雾霾图像的性质与分类研究进行了论述,并进一步综述了图像去雾技术的国内外研究现状,对直方图均衡化算法,Retinex算法和先验暗通道算法等典型的去雾方法的效果进行评价。总结了各类算法的性能,分析他们各自的优势和不足。最后指出了图像去雾技术的发展趋势和未来展望。     

退化图像复原方法研究进展

飞行器和空间成像制导装备在大气中高速飞行时会受到湍流干扰,导致光学系统接收到的图像发生模糊降质、像素偏移、信噪比降低等问题,开展退化图像复原技术及方法研究就成为空间光学成像系统获得较高性能图像的重要途径。通过对退化图像复原技术研究进展的系统梳理和分析研究,本文首先介绍了图像退化模型,接着给出了退化图像复原方法的分类,然后比较系统地介绍了确定正则化图像复原方法、随机正则化图像复原方法、基于局部相似性的图像复原方法、基于示例学习的图像复原方法等几种新型的单幅退化图像复原方法,其后分析了视频复原的特征、介绍了新近的几种典型的视频图像复原方法,最后分析总结出了图像复原的难点所在。对于促进我国退化图像复原技术的研究和发展具有一定的参考价值。     

A novel haze image steganography method via cover-source switching

In realistic outdoor scenarios, image sensors tend to suffer from various weather conditions (e.g., haze, rain, etc.),which make the images of the same scene taken at different times may be different. Therefore, one should be able to securely embed secret messages into these images by making use of the variations of the weather effects. Inspired by some recent natural steganography algorithms, this paper presents a novel haze image steganography method, which embeds messages through adjusting the weather effects of an input haze image, making it resemble the same image captured under another weather condition. The proposed steganography method consists of three parts: (1) model parameter estimation of the input haze image, (2) haze effects adjustment according to the atmospheric scattering model, (3) message embedding using the floating-point adjusted haze image. 10,000 haze images captured under different haze conditions in various scenarios were used to test the proposed steganography algorithm. The experimental results show that the proposed steganography algorithm is more secure than S-UNIWARD and HILL for steganalyzers who only have raw haze images.     

基于物理模型的快速单幅图像去雾方法

在雾天环境下,获取的图像受到大气粒子散射的影响而导致对比度和能见度降低.针对该问题,提出了一种基于物理模型的单幅图像的快速去雾方法.该方法以大气散射模型为基础,引入暗原色先验规律求取全局大气光,利用双边滤波局部估计雾浓度,间接求取大气耗散函数,最终通过变换的大气散射模型恢复无雾图像.大量实验结果表明,该方法能够恢复出自然清晰的无雾图像,并能够较好地处理景深突变的边缘及远景处.此外,该算法在处理图像的运算时间上具有明显的优势,可满足图像实时处理要求.     

基于水下成像模型的图像清晰化算法

受水下场景中有机物和悬浮颗粒的影响,水下图像存在对比度低、颜色失真和细节丢失等问题。同时,水下场景中通常有人工光源存在,造成图像光照不均。传统基于图像去雾的方法用于水下图像复原时效果欠佳,为充分考虑水对光的吸收和散射作用,近期提出了新的水下成像模型和图像复原方法。但是这些方法未考虑红通道影响,导致估计的散射比偏大;另外,也未考虑人工光源的影响,导致估计的背景光过大。针对这些问题,该文提出一套有效的水下图像清晰化方案。首先,通过设置阈值确定是否将红通道信息用于暗通道计算,并将反映人工光源影响的饱和度指标用于散射比估计,以减小人工光源的影响。由此,提出了基于红通道预判和饱和度指标的暗通道计算方法。然后,根据三通道衰减系数比估计每个通道的透射率,可弥补目前很多方法假设蓝绿通道透射率一致的缺陷。最后,利用Shades of Gray算法估计环境光,并结合新的水下成像模型得到复原图像。实验结果表明,该文算法可显著提升图像的对比度,得到颜色自然、细节清晰的复原图像。     

动态大气散射系数的颜色衰减先验图像去雾

针对颜色衰减先验图像去雾算法存在对较浓的有雾图像去雾效果不佳的问题,提出基于动态大气散射系数的颜色衰减先验图像去雾算法.用动态大气散射系数取代颜色衰减先验去雾算法中恒定大气散射系数的假设,定义大气散射系数为关于图像景深的指数函数.利用Middlebury stereo datasets中无雾图像和相应的景深图像得到合成有雾图像.采用均方误差(MSE)和结构相似度(SSIM)的综合评价参数MSE-SSIM确定上述指数函数的两个参数的最佳取值.实验结果表明与颜色衰减先验算法、He、Meng算法相比,该算法的去雾图像清晰颜色自然,有效地提高了去雾效果.     

Single underwater image haze removal with a learning-based approach to blurriness estimation

Underwater images are usually degraded due to light scattering and absorption. To recover the scene radiance of degraded underwater images, a new haze removal method is presented by incorporating a learning-based approach to blurriness estimation with the image formation model. Firstly, the image blurriness is estimated with a linear model trained on a set of selected grayscale images, the average Gaussian images and blurriness images. With the estimated image blurriness, three intermediate background lights (BLs) are computed to obtain the synthesized BL. Then the scene depth is calculated by using the estimated image blurriness and BL to construct a transmission map and restore the scene radiance. Compared with other haze removal methods, haze in degraded underwater images can be removed more accurately with our proposed method. Moreover, visual inspection, quantitative evaluation and application test demonstrate that our method is superior to the compared methods and beneficial to high-level vision tasks.     

图像去雾过程中的噪声抑制方法

大气中微小颗粒（如雾、霾等）的散射作用会使户外场景拍摄的图像发生退化,造成图像质量下降。图像去雾可以提升图像对比度,增加场景能见度,校正颜色失真,改善视觉效果。但是图像去雾经常会出现明显的噪声放大现象,尤其是无穷远处的天空区域最为严重。针对这一问题,提出了一种去雾过程中的噪声抑制方法。以传输率图像为指导,采用滤波半径变化的双边滤波对雾天图像进行模糊。再计算新的传输率图像,代入雾天成像模型,得到去噪后复原图像。结合噪声评价方法,实验结果验证了该方法的噪声抑制效果。     

A Model-based dehazing scheme for unmanned aerial vehicle system using radiance boundary constraint and graph model

Unmanned aerial vehicle system (UAVs) imaging has become a challenging area of research due to the dynamic atmospheric environment. The images captured by UAVs are often deteriorated by factors such as clouds occlusion, poor atmospheric illumination, and limited capability of the imaging system. To tackle problems, this paper presents a novel visibility restoration scheme for UAVs images by considering the following two assumptions: (1) The actual scene radiance of a UAVs image is bounded. (2) Pixels sharing the same appearance must have the same transmission value in a local neighborhood. Inspired by above assumptions, an image boundary constraint utilizing the median filter has been imposed on the RGB channel for the rough estimation of transmission-map in aerial images. Furthermore, a graph-model based optimization technique has been used for the transmission-map refinement. The experimental results demonstrate the efficiency of the proposed method in terms of metrics correspond to the human-visual-system (HVS).     

基于深度学习的图像去雾算法研究进展

图像是信息的重要承载形式。雾霾的出现降低了图像采集设备采集到的图像质量,容易出现色彩暗淡、对比度和饱和度降低、细节信息丢失等问题,直接影响了有用信息的表达和利用。目前对图像去雾的研究多采用深度学习的方法,卷积神经网络代替了人工特征提取方式,取得了优于传统算法的去雾效果,但普遍存在着对真实世界雾霾图像和清晰图像对的依赖。无监督学习的方法带来了新的解决思路。从监督学习和无监督学习的角度对有代表性的深度学习图像去雾算法进行分类,归纳了常用的数据集、评价指标,概括分析了有影响力的去雾模型的核心思想,总结了各算法的优缺点和适用场景。针对目前工作存在的不足,探索了下一步研究的方向。     

基于WLS的雾天交通图像恢复方法

在尘雾等恶劣天气条件下,由于大气粒子的散射作用,致使获取的道路图像严重退化,给交通运输带来很大的困难.为了提高道路环境的可视性,文中提出了一种基于WLS的雾天交通图像恢复算法.该算法从大气散射模型出发,首先进行大气光照的估计与白平衡处理,然后结合道路环境的约束,构建WLS框架对大气耗散函数进行估计,从而恢复场景反照率.通过实验分析可知,文中算法能够有效去除图像中雾霾,消除了Halo效应的影响,较好地凸显图像远景的细节信息,实现了交通图像的视见度的提高.     

Single image dehazing using a new color channel

Images with hazy scene suffer from low-contrast, which reduces the visible quality of the scene, thus making object detection a more challenging task. Low-contrast can result from foggy weather conditions during image acquisition. Dehazing is a process of removal of haze from the photography of a hazy scene. Single-image dehazing based on dark channel priors are well-known techniques in this field. However, the performance of such techniques is limited to priors or constraints. Moreover, this type of method fails when images have sky-region. So, a method is proposed, which can restore the visibility of hazy images. First, a hazy image is divided into blocks of size 32 × 32, then the score of each block is calculated to select a block having the highest score. Atmospheric light is calculated from the selected block. A new color channel is considered to remove atmospheric scattering, obtained channel value and atmospheric light are then used to calculate the transmission map in the second step. Third, radiance is computed using a transmission map and atmospheric light. The illumination scaling factor is adopted to enhance the quality of a dehazed image in the final step. Experiments are performed on six datasets namely, I-HAZE, O-HAZE, BSDS500, FRIDA, RESIDE dataset and natural images from Google. The proposed method is compared against 11 state-of-the-art methods. The performance is analyzed using fourteen quantitative evaluation metrics. All the results demonstrate that the proposed method outperforms 11 state-of-the-art methods in most of the cases.     

Research on the parallelization of image quality analysis algorithm based on deep learning

Image quality assessment is an indispensable in computer vision applications, such as image classification, image parsing. With the development of Internet, image data acquisition becomes more conveniently. However, image distortion is inevitable due to imperfect image acquisition system, image transmission medium and image recording equipment. Traditional image quality assessment algorithms only focus on low-level visual features such as color or texture, which could not encode high-level features effectively. CNN-based methods have shown satisfactory results in image quality assessment. However, existing methods have problems such as incomplete feature extraction, partial image block distortion, and inability to determine scores. So in this paper, we propose a novel framework for image quality assessment based on deep learning. We incorporate both low-level visual features and high-level semantic features to better describe images. And image quality is analyzed in a parallel processing mode. Experiments are conducted on LIVE and TID2008 datasets demonstrate the proposed model can predict the quality of the distorted image well, and both SROCC and PLCC can reach 0.92 or higher.