在雾天环境下所获取的图像会出现对比度下降、细节不清晰、局部亮度偏高、饱和度降低等一系列降质退化现象，所以去除雾天影响、复原雾天图像，在视频监测、无人驾驶等领域具有重要意义。目前大部分雾天图像复原算法的应用场景都是户外白天，但夜间场景由于人造光源的加入更加复杂，白天复原算法在夜间场景下复原效果下降甚至失效。

针对现有雾天图像复原算法其结果在白天场景中易出现天空失真、细节模糊以及部分图像整体亮度偏暗等问题，基于暗通道先验算法进行改进，提出了一种基于天空区域判别分割的雾天图像复原算法。根据图像天空区域亮度高、相对平滑的特点，利用亮度以及梯度阈值对图像天空区域进行判别分割。对含有天空区域的雾图，用天空区域平均像素值作为大气光值，增强了大气光值选取的合理性；通过小波变换以及分割矩阵将亮度模型透射率以及暗通道透射率进行融合，从而保证整体图像透射率估计准确。对于不含天空区域的雾图，以暗通道算法为基础，根据输入图像大小确定最小值滤波窗口大尺寸，避免复原结果细节模糊问题。

针对夜间环境下雾天图像复原算法其结果易出现的光源区域细节损失、光晕扩大以及色彩失真的问题，通过对基于亮暗通道融合的夜间图像复原算法进行改进，提出了一种基于光源区域分割以及透射率补偿的夜间雾图复原算法。该算法根据光源的空间特性，采用引导滤波以及亮通道图对大气光图进行全局估计；通过设置自动阈值，利用雾天图像颜色通道差异进行光源区域分割；同时根据光源区域与非光源区域的透射率差异以及夜间图像整体偏暗的特性，以暗通道透射率为基础，亮度图为表征，利用亮通道透射率对光源区域透射率补偿并通过块操作进行色偏处理。

实验结果表明，基于天空区域判别分割的白天雾图复原算法对大气光值以及透射率图的求取较为准确，得到的复原图像清晰自然，天空区域未出现失真等现象，复原效果良好，符合人眼视觉效果；在客观评价指标峰值信噪比、结构相似性指标上提升明显，图像信息熵保持良好，对于白天不同场景下雾天图象处理具有普遍适用性。基于透射率补偿的夜间雾图复原算法对图像的光源区域分割准确，针对透射率以及大气光值的求取符合夜间图像空间特性，复原图像色调自然，同时避免了光源扩散等问题，具有良好的复原效果；较之对比算法，客观评价指标峰值信噪比、结构相似性以及信息熵都有所提升，结合主观视觉评价，该方法适用于夜间雾图复原，得到的复原图像具有较高的清晰度。

The images acquired in foggy environment will show a series of degradation phenomena, such as contrast reduction, unclear details, high local brightness, and reduced saturation. Therefore, removing the influence of foggy weather and restoring foggy images are of great significance in video monitoring, unmanned driving and other fields. At present, most of the application scenarios of image dehazing algorithms are outdoor during the day. However, due to the addition of artificial light sources, the nighttime scene is more complex, and the daytime dehazing algorithm is ineffective in the nighttime scene.

Aiming at the problems of sky distortion, blurred details and dark overall brightness of some images in the daytime scene of the existing foggy image restoration algorithm, in thesis we improved the dark channel prior algorithm and proposed a foggy image restoration algorithm based on sky region discriminative segmentation. According to the characteristics of high brightness and relative smoothness of the sky region of the image, the sky region of the image is discriminated and segmented by using brightness and gradient threshold. For the fog map with sky region, the average pixel value of the sky region was used as the atmospheric light value, which enhanced the rationality of the atmospheric light value selection. The brightness model transmittance and dark channel transmittance are fused by wavelet transform and segmentation matrix to ensure the accuracy of the overall image transmittance estimation. For the fog image without sky area, based on the dark channel algorithm, the size of the minimum filtering window is determined according to the size of the input image to avoid the problem of blurred details of the restoration result.

In order to solve the problems of light source region detail loss, halo enlargement and color distortion, a night fog image restoration algorithm based on light source region segmentation and transmission compensation is proposed by improving the night image restoration algorithm based on light and dark channel fusion. According to the spatial characteristics of the light source, the algorithm uses guided filtering and bright channel diagram to estimate the atmospheric light map globally. By setting automatic threshold, the light source area is segmented according to the color channel difference of fog image. At the same time, according to the transmittance difference between the light source area and the non-light source area and the overall dark characteristics of the night image, the brightness map is characterized based on the dark channel transmittance. The bright channel transmittance is used to compensate the transmittance of the light source area and the color bias is processed by block operation.

The experimental results show that the image dehazing algorithm based on sky region discrimination segmentation is more accurate in obtaining the atmospheric light value and transmission map, and the restored image is clear and natural. There is no distortion in the sky region, and the dehazing effect is good, which is in line with the human visual effect. The peak signal to noise ratio (PSNR) and structural similarity index (SSIM) are improved significantly, and the image information entropy is maintained well. It has universal applicability for foggy image processing in different scenes during the day. The night image dehazing algorithm based on transmittance compensation can segment the light source area of the image accurately. The calculation of transmittance and atmospheric light value is consistent with the spatial characteristics of the night image, the restored image is natural in tone, and the problems such as light source diffusion are avoided, which has a good dehazing effect. Compared with the comparison algorithm, the objective evaluation indicators peak signal-to-noise ratio, structural similarity and information entropy are improved. Combined with subjective visual evaluation, the proposed method is suitable for night dehazing, and the restored image has higher definition.

[1]Vijayalakshmi D, Nath M K. A novel multilevel framework based contrast enhancement for uniform and non-uniform background images using a suitable histogram equalization[J]. Digital Signal Processing, 2022, 127(12): 103532-103544.

[2]Paul A, Bhattacharya P, Maity S P. Histogram modification in adaptive bi-histogram equalization for contrast enhancement on digital images[J]. Optik, 2022, 259: 168899.

[3]Paul A. Adaptive tri-plateau limit tri-histogram equalization algorithm for digital image enhancement[J]. The Visual Computer, 2023, 39(1): 297-318.

[4]Li C L, Liu J H, Liu A Y, et al. Global and Adaptive Contrast Enhancement for Low Illumination Gray Images[J]. IEEE Access, 2019，7: 163395–163411.

[5]Stimper V, Bauer S, Ernstorfer R, et al. Multidimensional Contrast Limited Adaptive Histogram Equalization[J]. IEEE Access, 2019, 7: 165437–165447.

[6]Liu X, Zhang H, Cheung Y, et al. Efficient single image dehazing and denoising: An efficient multi-scale correlated wavelet approach[J]. Computer Vision and Image Understanding, 2017, 162(4): 23-33.

[7]Khmag A, Al-Haddad S A R, Ramli A R, et al. Single image dehazing using second-generation wavelet transforms and the mean vector L2-norm[J]. The visual computer, 2018, 34(2): 675-688.

[8]韦春苗，徐岩，李媛．基于小波变换的迭代融合去雾算法[J]．激光与光电子学进展，2021，58(20)：243-251．

[9]李婵飞．基于图像融合的单幅图像去雾方法[J]．广东通信技术，2022，42(03)：77-79．

[10]Jobson D J, Rahman Z-u, Woodell G A. A multiscale retinex for bridging the gap between color images and the human observation of scenes [J]. IEEE Transactions on Image processing, 1997,6 (7): 965–976.

[11]李旺，杨金宝，孙婷，等．基于Retinex的多尺度单幅图像去雾网络[J]．青岛大学学报（自然科学版），2022，35(04)：26-32．

[12]李竹林，李鹏翼，车雯雯．一种基于Retinex理论的图像去雾算法研究[J]．河南科学，2023，41(01)：1-6．

[13]吴向平，高庆庆，黄少伟，等．基于景深信息的自适应Retinex图像去雾算法[J]．激光与光电子学进展，2023，60(12)：160-169．

[14]Meng G, Wang Y, Duan J, et al. Efficient image dehazing with boundary constraint and contextual regularization[C]//2013 IEEE International Conference on Computer Vision, December 1-8, 2013, Sydney, NSW, Australia. New York: IEEE, 2013: 617-624.

[15]Zhu Q, Mai J, Shao L. A fast single image haze removal algorithm using color attenuation prior[J]. IEEE transactions on image processing, 2015, 24(11): 3522-3533.

[16]He K, Sun J, Tang X. Single image haze removal using dark channel prior[J]. IEEE transactions on pattern analysis and machine intelligence, 2010, 33(12): 2341-2353.

[17]谢昊伶，彭国华，王凡．基于背景光估计与暗通道先验的水下图像复原[J]．光学学报，2018，38(01)：18-27．

[18]梅英杰，宁媛，陈进军．融合暗通道先验和MSRCR的分块调节图像增强算法[J]．光子学报，2019，48(07)：124-135．

[19]Mao T, Wang J, Sun Z, et al. Visual Optimizing Technology of Dark Channel Prior Dehazing Based on Sky Region Segmentation[J]. Semiconductor Optoelectronics, 2017, 38(6):902-907.

[20]Wang W, Lian X, Wu X, et al. Fast image dehazing method based on linear transformation[J]. IEEE Transactions on Multimedia, 2017, 19(6): 1142-1155.

[21]Salazar-Colores S, Cabal-Yepez E, Ramos-Arreguin J M, et al. A fast image dehazing algorithm using morphological reconstruction[J]. IEEE Transactions on Image Processing, 2018, 28(5): 2357-2366.

[22]Gao Y, Hu H M, Li B, et al. Detail preserved single image dehazing algorithm based on airlight refinement[J]. IEEE Transactions on Multimedia, 2018, 21(2): 351-362.

[23]Kim S E, Park T H, Eom I K. Fast Single Image Dehazing Using Saturation Based Transmission Map Estimation[J]. IEEE Transactions on Image Processing, 2019, 29: 1985-1998.

[24]Zhao D, Xu L, Yan Y, et al. Multiscale optimal fusion model for single image dehazing[J]. Signal Processing: Image Communication, 2019, 74: 253-265.

[25]Li S, Zhou Q H. Single image dehazing based on fusion of sky region segmentation[J]. Journal of Physics: Conference Series, 2021, 1971(1): 012093

[26]黄鹤，李战一，胡凯益，等．融合大气光值-图估计的无人机航拍图像去雾[J]．哈尔滨工业大学学报，2023，55(05)：88-97．

[27]陆松浩．单幅图像去雾算法研究[D]．南京：南京邮电大学，2023．

[28]仲会娟，马秀荣，张静怡，等．基于超像素暗通道和自动色阶优化的图像去雾算法[J]．光电子·激光，2023，34(10)：1059-1067．

[29]吴曙镔，于万钧，陈颖．基于暗亮通道先验的小波融合图像去雾算法[J/OL]．电光与控制，1-8[2024-05-28]．

[30]石冬阳，张俊林，刘天光，等．基于容差机制与高斯滤波的图像去雾算法[J]．重庆科技学院学报（自然科学版），2023，25(05)：56-62．

[31]Cai B, Xu X, Jia K, et al. DehazeNet: an end-to-end system for single image haze removal[J]. IEEE Transactions on Image Processing, 2016, 25(11): 5187-5198.

[32]Ren W Q, Liu S, Zhang H, et al. Single image dehazing via multi-scale convolutional neural networks[C]. European Conference on Computer Vision, 2016: 154-169.

[33]Ren W, Ma L, Zhang J, et al. Gated fusion network for single image dehazing[C]. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018: 3253–3261.

[34]H. Dong, J. Pan, L. Xiang, et al. Multi-scale boosted dehazing network with dense feature fusion[C]. in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020: 2157–2167.

[35]Li B, Peng X, Wang Z, et al. AOD-Net: All-in-One Dehazing Network[C]. IEEE International Conference on Computer Vision, 2017: 4770-4778.

[36]Zhang H, Patel V M. Densely connected pyramid dehazing network[C]. IEEE Conference on Computer Vision and Pattern Recognition, 2018. 3194-3203.

[37]徐祯东，张天宇，张世恒等．基于YUV颜色空间GAN网络的图像去雾算法研究[J]．图学学报，2023，44(05)：928-936．

[38]Chen D, He M, Fan Q, et al. Gated Context Aggregation Network for Image Dehazing and Deraining[C]. IEEE Winter Conference on Applications of Computer Vision, 2018:1550-5790.

[39]Ren W Q, Ma L, Zhang J W, et al. Gated fusion network or single image dehazing[C]. IEEE Conference on Computer Vision and Pattern Recognition, 2018: 3253-3261.

[40]Qu Y Y, Chen Y Z, Huang J Y, et al. Enhanced pix2pix dehazing network[C]. IEEE Conference on Computer Vision and Pattern Recognition, 2019: 8160-8168.

[41]Qin X, Wang Z L, Bai Y C, et al. FFA-Net: Feature fusion attention network for single image dehazing[C]. Association for the Advance of Artificial Intelligence, 2020: 11908-11915.

[42]Guo F, Zhao X, Tang J, et al. Single image dehazing based on fusion strategy[J]. Neurocomputing, 2020, 378: 9-23.

[43]Yang H H, Yang C H, Tsai Y C J. Y-net: Multi-scale feature aggregation network with wavelet structure similarity loss function for single image dehazing[C]. IEEE International Conference on Acoustics, 2020: 2628-2632.

[44]朱伟，段跳楠，吉咸阳，等．基于轻量化深度神经网络（LDNet）的图像去雾方法[J]．指挥信息系统与技术，2023，14(05)：86-93．

[45]许广峰．基于特征金字塔网络的轻量化图像去雾算法及模型优化方法研究[D]．南京：南京邮电大学，2023．

[46]刘致远，但志平．基于注意力增强的CycleGAN图像去雾[J]．国外电子测量技术，2023，42(09)：162-168．

[47]Pei S C, Lee T Y. Nighttime haze removal using color transfer pre-processing and dark channel prior[C]. Proceedings of 2012 19th IEEE International Conference on Image Processing, IEEE, 2012.

[48]Zhang J, Cao Y, Wang Z F. Nighttime haze removal based on a new imaging model[C]. Proceedings of 2014 IEEE International Conference on Image Processing, IEEE, 2014.

[49]Li Y, Tan R T, Brown M S. Nighttime haze removal with glow and multiple light colors[C]. Proceedings of 2015 IEEE International Conference on Computer Vision, IEEE, 2015.

[50]Santra S，Chanda B. Day/night unconstrained image dehazing [C]//Proceedings of 2016 23rd International Conference on Pattern Recognition. Cancun，Mexico: IEEE，2016: 1406-1411.

[51]Ancuti C, Ancuti C, Vleeschouwer C, et al. Nighttime dehazing by fusion[C]. IEEE International Conference on Image Processing(ICIP), Phoenix, USA , 2016: 2256-2260.

[52]Li Z G, Wei Z, Wen C Y, et al. Detail enhanced multi-Scale exposure fusion[J]. IEEE Transactions on Image Processing, 2017, 26(3): 1243-1252.

[53]Yu T, Song K, Miao P, et al. Nighttime single image dehazing via pixel-wise alpha blending[J]. IEEE Access, 2019,7: 114619-114630.

[54]杨爱萍，白煌煌. 基于Retinex理论和暗通道先验的夜间图像去雾算法[J]. 激光与光电子学进展，2017，54（4）：147-153.

[55]Yang M M, Liu J C, Li Z G. Superpixel-based single nighttime image haze removal[J]. IEEE Transactions on Multimedia, 2018, 20(11): 3008-3018.

[56]陈志恒，严利民，张竞阳．采用自适应全局亮度补偿的夜间去雾算法[J]．红外技术，2021，43(10)：954-959．

[57]王同森，史勤忠，王得法，等．基于光源区域自适应的夜间去雾方法[J]．计算机科学，2021，48(S2)：327-333．

[58]薛楠，严利民．一种改进的透射率分布估计的夜间图像去雾算法[J]．红外技术，2022，44(10)：1089-1094．

[59]Mundy W C, Roux J A, Smith A M. Mie scattering by spheres in an absorbing medium [J]. Journal of the Optical Society of America, 1974, 64(12): 1593-1597.

[60]Li B, Ren W, Fu D, et al. Benchmarking single-image dehazing and beyond[J]. IEEE Transactions on Image Processing, 2018, 28(1): 492-505.

[61]Li Y, You S, Brown M S, et al. Haze visibility enhancement: A survey and quantitative benchmarking[J]. Computer Vision and Image Understanding, 2017, 165: 1-16.

[62]Kang L, Ye P, Li Y, et al. Convolutional neural networks for no-reference image quality assessment[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2014: 1733-1740.

[63]Jacobson, R. E. Image Quality Metrics [J]. The Journal of Photographic Science, 1995, 43(2): 42-43.

[64]Sandić-Stanković D, Kukolj D, Callet P L. Image quality assessment based on pyramid decomposition and mean squared error [M]. Telecommunications Forum Telfor. 2015.

[65]郭璠，蔡自兴．图像去雾算法清晰化效果客观评价方法[J]．自动化学报，2012, 38(09)：1410-1419．

[66]卜丽静，王涛．基于HVS的SSIM超分辨率重建图像质量评价方法[J]．测绘与空间地理信息，2019，42(7)：14-18+21．

[67]吕建威，钱锋，韩昊男，等．结合光源分割和线性图像深度估计的夜间图像去雾[J]．中国光学，2022，15(01)：34-44．

[68]张振海，贾争满，季坤．基于改进的Otsu法的地铁隧道裂缝识别方法研究[J]．重庆交通大学学报：自然科学版，2022，41(1)：84-90．

[69]Wang S, Zheng H J, Hu H M, et al. Naturalness preserved enhancement algorithm for non-uniform illumination images[J]. IEEE Transactions on Image Processing. 2013, 22(9), 3538-3548.