-
摘要:
针对夜间有雾图像受人工光源影响，存在亮度低、辉光及色偏问题，提出一种基于非均匀大气光修正模型的夜间图像去雾算法。结合非均匀大气光成像特性，通过加入表面照射光和辉光项修正传统大气散射模型；根据图像亮度和饱和度分量引入近光源系数，准确估计辉光项以校正非均匀大气光；基于暗通道先验(DCP)及带色彩恢复的多尺度Retinex增强算法求解模型，使用引导滤波解决透射率块状效应，采用线性融合方法解决过度增强问题；利用高斯高通滤波增强图像细节，得到最终的清晰图像。实验结果表明：所提算法恢复的夜间图像颜色自然、细节丰富且可见度高；相比暗通道先验等经典算法，以及联合对比度增强和曝光融合(CEEF)等新颖算法，色差降低约15%，峰值信噪比提升约15%，平均梯度提升约8%，在定性和定量实验中更具优势。
Abstract:To address the problems of low brightness, glow, and color distortion caused by artificial light sources in nighttime hazy images, this paper proposed a nighttime image dehazing algorithm based on a non-uniform atmospheric light correction model. Firstly, according to the imaging characteristics of non-uniform atmospheric light, the traditional atmospheric scattering model was modified by adding surface illumination light and glow terms. Secondly, the near-light source coefficient was determined based on image intensity and saturation components, and the glow term was accurately estimated to correct the non-uniform atmospheric light. Then, based on the dark channel prior (DCP) and multi-scale Retinex enhancement algorithm with color restoration ability, the study adopted guided filtering to solve the block effect of transmittance and the linear fusion to solve the problem of excessive enhancement. Finally, Gaussian high-pass filter was used for enhancing the details of the image to obtain the final clear image. The experimental results show that nighttime hazy images restored by the proposed algorithm have natural color, rich details, and high visibility. Compared to classical algorithms such as dark channel prior, as well as novel algorithms like contrast enhancement and exposure fusion (CEEF), the proposed algorithm reduces the color difference by about 15%, improves the peak signal-to-noise ratio by about 15%, and enhances the average gradient by about 8%, having more advantages in qualitative and quantitative experiments.
-
图 9 人工合成夜间有雾图像去雾结果对比
Figure 9. Comparison of dehazing results of synthetic nighttime hazy images
图 10 真实夜间有雾图像去雾结果对比
Figure 10. Comparison of dehazing results of real nighttime hazy images
图 11 图9中4幅图像的有参考指标曲线
Figure 11. Curves of two reference indexes of four images in Fig. 9
图 12 图9和图10中12幅图像的无参考指标曲线
Figure 12. Curves of three non-reference indexes of twelve images in Fig. 9 and Fig. 10
图 14 排除失真和伪影因素影响前后图像IE值对比
Figure 14. Comparison of image IE before and after eliminating distortion and artifact
表 1 CIEDE2000评测值对比
Table 1. Comparison of CIEDE2000 metrics
算法 色差值 色差值均值 黄色 白色 褐色 红色 蓝色 绿色 He2011[7] 10.384 14.362 12.356 13.031 33.204 19.878 17.203 Zhang2014[11] 8.543 11.497 10.409 24.261 30.797 12.618 16.354 Li2015[12] 6.604 4.915 16.213 9.109 32.103 9.080 13.004 OSFD2020[15] 3.800 3.637 12.518 13.427 40.042 11.388 14.135 CEEF2022[6] 26.813 26.068 15.886 9.713 40.018 31.475 24.996 本文算法 1.955 2.321 9.936 10.780 31.684 9.297 10.996 
下载: 导出CSV
表 2 人工合成夜间有雾图像数据集SSIM和PSNR均值
Table 2. Average values of SSIM and PSNR on artificial synthetic nighttime haze image dataset
下载: 导出CSV
表 5 消融实验的客观评价指标
Table 5. Objective evaluation indexes of ablation experiment
方法 AG IE NIQE 原始有雾图像 4.426 7.069 2.786 未去除表面照射光 5.199 6.513 2.791 未进行融合和增强 7.100 7.316 2.266 未去除辉光 10.692 7.373 2.109 本文算法 11.034 7.381 2.073
下载: 导出CSV
-
[1] 杨勇, 邱根莹, 黄淑英, 等. 基于改进大气散射模型的单幅图像去雾方法[J]. 北京麻豆精品秘 国产传媒学报, 2022, 48(8): 1364-1375.YANG Y, QIU G Y, HUANG S Y, et al. Single image dehazing method based on improved atmospheric scattering model[J]. Journal of Beijing University of Aeronautics and Astronautics, 2022, 48(8): 1364-1375(in Chinese). [2] 杨燕, 张得欣, 岳辉. 基于最小值通道与对数衰减的图像融合去雾算法[J]. 北京麻豆精品秘 国产传媒学报, 2020, 46(10): 1844-1852.YANG Y, ZHANG D X, YUE H. Image fusion dehazing algorithm based on minimum channel and logarithmic attenuation[J]. Journal of Beijing University of Aeronautics and Astronautics, 2020, 46(10): 1844-1852(in Chinese). [3] PAUL A. Adaptive tri-plateau limit tri-histogram equalization algorithm for digital image enhancement[J]. The Visual Computer, 2023, 39(1): 297-318. doi: 10.1007/s00371-021-02330-z [4] FU H Y, ZHENG W K, MENG X Y, et al. You do not need additional priors or regularizers in Retinex-based low-light image enhancement[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE Press, 2023: 18125-18134. [5] 刘备, 边少锋, 纪兵, 等. 小波变换在船载海洋重力测量数据处理与分析中的应用[J]. 系统工程与电子技术, 2023, 45(3): 654-659.LIU B, BIAN S F, JI B, et al. Application of wavelet transform in shipborne ocean gravimetry data processing and analysis[J]. Systems Engineering and Electronics, 2023, 45(3): 654-659(in Chinese). [6] LIU X N, LI H, ZHU C. Joint contrast enhancement and exposure fusion for real-world image dehazing[J]. IEEE Transactions on Multimedia, 2022, 24: 3934-3946. [7] HE K M, SUN J, TANG X O. Single image haze removal using dark channel prior[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2011, 33(12): 2341-2353. doi: 10.1109/TPAMI.2010.168 [8] LI C, YUAN C J, PAN H B, et al. Single-image dehazing based on improved bright channel prior and dark channel prior[J]. Electronics, 2023, 12(2): 299. doi: 10.3390/electronics12020299 [9] CUI G M, MA Q, ZHAO J F, et al. Image dehazing algorithm based on optimized dark channel and haze-line priors of adaptive sky segmentation[J]. Journal of the Optical Society of America A, 2023, 40(6): 1165-1182. doi: 10.1364/JOSAA.484423 [10] PEI S C, LEE T Y. Nighttime haze removal using color transfer pre-processing and dark channel prior[C]//Proceedings of the 19th IEEE International Conference on Image Processing. Piscataway: IEEE Press, 2012: 957-960. [11] ZHANG J, CAO Y, WANG Z F. Nighttime haze removal based on a new imaging model[C]//Proceedings of the IEEE International Conference on Image Processing. Piscataway: IEEE Press, 2014: 4557-4561. [12] LI Y, TAN R T, BROWN M S. Nighttime haze removal with glow and multiple light colors[C]//Proceedings of the IEEE International Conference on Computer Vision. Piscataway: IEEE Press, 2015: 226-234. [13] LIAO Y H, SU Z, LIANG X G, et al. HDP-Net: haze density prediction network for nighttime dehazing[C]//Proceedings of the Pacific Rim Conference on Multimedia. Berlin: Springer, 2018: 469-480. [14] KUANAR S, RAO K R, MAHAPATRA D, et al. Night time haze and glow removal using deep dilated convolutional network[EB/OL]. (2019-02-03)[2023-07-01]. http://arxiv.org/abs/1902.00855v1. [15] ZHANG J, CAO Y, ZHA Z J, et al. Nighttime dehazing with a synthetic benchmark[C]//Proceedings of the 28th ACM International Conference on Multimedia. New York: ACM, 2020: 2355-2363. [16] 鲁丹. 不均匀光照下单幅雾天图像复原算法研究[D]. 合肥: 合肥工业大学, 2015: 17-44.LU D. Research on restoration algorithm of single foggy image under uneven illumination[D]. Hefei: Hefei University of Technology, 2015: 17-44(in Chinese). [17] 何弘. 非均匀大气光视频图像去雾算法及其FPGA实现[D]. 长沙: 湖南大学, 2021: 14-16.HE H. Defogging algorithm for non-uniform atmospheric light video image and its FPGA implementation[D]. Changsha: Hunan University, 2021: 14-16(in Chinese). [18] HE Y F, LI C L, LI X. Remote sensing image dehazing using heterogeneous atmospheric light prior[J]. IEEE Access, 2023, 11: 18805-18820. doi: 10.1109/ACCESS.2023.3247967 [19] LAI Y, LIU Y. Improved single image dehazing with heterogeneous atmospheric light estimation[C]//Proceedings of the 2nd CCF Chinese Conference. Berlin: Springer, 2017: 102-112. [20] 赖欣, 王储, 陈航. 低照度下人脸检测MSRCR光频分段滤波增强算法[J]. 电子测量与仪器学报, 2022, 36(2): 96-106.LAI X, WANG C, CHEN H. MSRCR optical frequency segmented filter enhancement algorithm in low-light face detection[J]. Journal of Electronic Measurement and Instrumentation, 2022, 36(2): 96-106(in Chinese). [21] 杨燕, 张金龙, 张浩文. 基于区间估计与透射率自适应约束的去雾算法[J]. 北京麻豆精品秘 国产传媒学报, 2022, 48(1): 15-26.YANG Y, ZHANG J L, ZHANG H W. Dehazing algorithm based on interval estimation and adaptive constraints of transmittance[J]. Journal of Beijing University of Aeronautics and Astronautics, 2022, 48(1): 15-26(in Chinese). [22] HE K M, SUN J, TANG X O. Guided image filtering[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2013, 35(6): 1397-1409. doi: 10.1109/TPAMI.2012.213 [23] ZHENG Z R, REN W Q, CAO X C, et al. Ultra-high-definition image dehazing via multi-guided bilateral learning[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE Press, 2021: 16180-16189. [24] LI B Y, REN W Q, FU D P, et al. Benchmarking single-image dehazing and beyond[J]. IEEE Transactions on Image Processing, 2019, 28(1): 492-505. doi: 10.1109/TIP.2018.2867951 [25] SHARMA G, WU W, DALAL E N. The CIEDE2000 color-difference formula: implementation notes, supplementary test data, and mathematical observations[J]. Color Research and Application, 2005, 30(1): 21-30. [26] 杨紫骞, 王艳秋, 郑福, 等. 多维点云结构相似性定量化评价[J]. 光学精密工程, 2023, 31(4): 533-542. doi: 10.37188/OPE.20233104.0533YANG Z Q, WANG Y Q, ZHENG F, et al. Quantitative evaluation method for structural similarity of multidimensional point cloud[J]. Optics and Precision Engineering, 2023, 31(4): 533-542(in Chinese). doi: 10.37188/OPE.20233104.0533 [27] ZHAO B, LIU Z Y, DING S X, et al. Motion artifact correction for MR images based on convolutional neural network[J]. Optoelectronics Letters, 2022, 18(1): 54-58. doi: 10.1007/s11801-022-1084-z [28] 王殿伟, 韩鹏飞, 范九伦, 等. 基于光照-反射成像模型和形态学操作的多谱段图像增强算法[J]. 物理学报, 2018, 67(21): 104-114.WANG D W, HAN P F, FAN J L, et al. Multispectral image enhancement based on illuminance-reflection imaging model and morphology operation[J]. Acta Physica Sinica, 2018, 67(21): 104-114(in Chinese). [29] 罗国强. Gamma变换与多尺度细节增强的红外图像增强[J]. 激光与红外, 2023, 53(2): 253-260.LUO G Q. Infrared image enhancement based on Gamma transformation and multi-scale detail enhancement[J]. Laser & Infrared, 2023, 53(2): 253-260(in Chinese). -
下载:
点击查看大图
计量
- 文章访问数: 207
- HTML全文浏览量: 86
- PDF下载量: 9
- 被引次数: 0
