2.845

2023影响因子

(CJCR)

  • 中文核心
  • EI
  • 中国科技核心
  • Scopus
  • CSCD
  • 英国科学文摘

留言板

尊敬的读者、作者、审稿人, 关于本刊的投稿、审稿、编辑和出版的任何问题, 您可以本页添加留言。我们将尽快给您答复。谢谢您的支持!

姓名
邮箱
手机号码
标题
留言内容
验证码

一种基于改进AOD-Net的航拍图像去雾算法

李永福 崔恒奇 朱浩 张开碧

李永福, 崔恒奇, 朱浩, 张开碧. 一种基于改进AOD-Net的航拍图像去雾算法. 自动化学报, 2022, 48(6): 1543−1559 doi: 10.16383/j.aas.c210232
引用本文: 李永福, 崔恒奇, 朱浩, 张开碧. 一种基于改进AOD-Net的航拍图像去雾算法. 自动化学报, 2022, 48(6): 1543−1559 doi: 10.16383/j.aas.c210232
Li Yong-Fu, Cui Heng-Qi, Zhu Hao, Zhang Kai-Bi. A defogging algorithm for aerial image with improved AOD-Net. Acta Automatica Sinica, 2022, 48(6): 1543−1559 doi: 10.16383/j.aas.c210232
Citation: Li Yong-Fu, Cui Heng-Qi, Zhu Hao, Zhang Kai-Bi. A defogging algorithm for aerial image with improved AOD-Net. Acta Automatica Sinica, 2022, 48(6): 1543−1559 doi: 10.16383/j.aas.c210232

一种基于改进AOD-Net的航拍图像去雾算法

doi: 10.16383/j.aas.c210232
基金项目: 国家自然科学基金(U1964202, 61773082, 62073052), 重庆市自然科学基金(cstc2021jcyj-msxmX0373), 重庆邮电大学基金(A2018-02)资助
详细信息
    作者简介:

    李永福:重庆邮电大学自动化学院教授. 主要研究方向为智能网联汽车, 空地协同控制. 本文通信作者. E-mail: liyongfu@cqupt.deu.cn

    崔恒奇:重庆邮电大学硕士研究生. 主要研究方向为航拍图像处理. E-mail: cuihengqi2020@163.com

    朱浩:重庆邮电大学自动化学院教授. 主要研究方向为智能车环境感知与信息融合. E-mail: zhuhao@cqupt.edu.cn

    张开碧:重庆邮电大学自动化学院副教授. 主要研究方向为计算机应用与图像处理. E-mail: zhangkb@cqupt.edu.cn

A Defogging Algorithm for Aerial Image With Improved AOD-Net

Funds: Supported by National Natural Science Foundation of China (U1964202, 61773082, 62073052), Chongqing Natural Science Foundation (cstc2021jcyj-msxmX0373), and Chongqing University of Posts and Telecommunications Foundation (A2018-02)
More Information
    Author Bio:

    LI Yong-Fu Professor at the College of Automation, Chongqing University of Posts and Telecommunications. His research interest covers connected and automated vehicles and airground cooperative control. Corresponding author of this paper

    CUI Heng-Qi Master student at the College of Automation, Chongqing University of Posts and Telecommunications. His research interest covers aerial image processing

    ZHU Hao Professor at the College of Automation, Chongqing University of Posts and Telecommunications. His research interest covers environmental perception of intelligent vehicles and information fusion

    ZHANG Kai-Bi Associate professor at the College of Automation, Chongqing University of Posts and Telecommunications. Her research interest covers computer application and image processing

  • 摘要: 针对航拍图像易受雾气影响, AOD-Net (All in one dehazing network)算法对图像去雾后容易出现细节模糊、对比度过高和图像偏暗等问题, 本文提出了一种基于改进AOD-Net的航拍图像去雾算法. 本文主要从网络结构、损失函数、训练方式三个方面对AOD-Net进行改良. 首先在AOD-Net的第二个特征融合层上添加了第一层的特征图, 用全逐点卷积替换了传统卷积方式, 并用多尺度结构提升了网络对细节的处理能力. 然后用包含有图像重构损失函数、SSIM (Structural similarity)损失函数以及TV (Total variation)损失函数的复合损失函数优化去雾图的对比度、亮度以及色彩饱和度. 最后采用分段式的训练方式进一步提升了去雾图的质量. 实验结果表明, 经该算法去雾后的图像拥有令人满意的去雾结果, 图像的饱和度和对比度相较于AOD-Net更自然. 与其他对比算法相比, 该算法在合成图像实验、真实航拍图像实验以及算法耗时测试的综合表现上更好, 更适用于航拍图像实时去雾.
  • 图  1  AOD-Net的网络结构

    Fig.  1  The network architecture of AOD-Net

    图  2  本文所提的网络结构

    Fig.  2  The proposed network architecture

    图  3  去雾效果 ((a) 有雾图像; (b) AOD-Net;(c) 改良后的AOD-Net)

    Fig.  3  Defogging effect ((a) Fog image; (b) AOD-Net; (c) Improved AOD-Net)

    图  4  多尺度网络结构

    Fig.  4  The architecture of multi-scale network

    图  5  不同方法训练本文所提模型获得的损失曲线 ((a) 单一函数训练方法; (b) 分段函数训练方法)

    Fig.  5  The loss curve obtained by training the proposed model with different methods ((a) The training method of single function; (b) The training method of piecewise function)

    图  6  不同训练方式下的SSIM与PSNR变化曲线((a) SSIM曲线; (b) PSNR曲线)

    Fig.  6  The curve of SSIM and PSNR under different training methods ((a) The curve of SSIM; (b) The curve of PSNR)

    图  7  两种不同训练方法下的去雾效果 ((a) 合成雾图; (b) ground truth; (c) 所提模型用LMS训练1000次的效果; (d) 所提模型用LMS训练1500次的效果; (e) 所提模型用式L训练1000次的效果; (f) 所提模型用式L训练1500次的效果)

    Fig.  7  Defogging effect of two different training method ((a) Synthetic fog image; (b) Ground truth; (c) The proposed model was trained after 1000 times by LMS; (d) The proposed model was trained after 1500 times by LMS; (e) The proposed model was trained after 1000 times by L; (f) The proposed model was trained after 1500 times by L)

    图  8  合成有雾图像的实验结果展示 ((a) 有雾图像; (b) Ground truth; (c) DCP; (d) BCCR; (e) CAP; (f) DehazeNet;(g) MSCNN; (h) AOD-Net; (i) GFN; (j) GCANet; (k) FFANet; (l) 本文算法)

    Fig.  8  Experimental results of the synthetic fog images ((a) Fog image; (b) Ground truth; (c) DCP; (d) BCCR; (e) CAP; (f) DehazeNet; (g) MSCNN; (h) AOD-Net; (i) GFN; (j) GCANet; (k) FFANet; (l) Proposed algorithm)

    图  9  真实有雾航拍图像的实验结果展示 ((a) 有雾图像; (b) DCP; (c) BCCR; (d) CAP; (e) DehazeNet; (f) MSCNN;(g) AOD-Net; (h) GFN; (i) GCANet; (j) FFANet; (k) 本文算法)

    Fig.  9  Experimental results of the real aerial fog images ((a) Aerial fog image; (b) DCP; (c) BCCR; (d) CAP; (e) DehazeNet; (f) MSCNN; (g) AOD-Net; (h) GFN; (i) GCANet; (j) FFANet; (k) Proposed algorithm)

    图  10  不同去雾算法在真实航拍有雾图像上的客观评价结果 ((a) BIQME; (b) FADE; (c) VI;(d) RI; (e) CB; (f) VIF; (g) GB; (h) Entropy)

    Fig.  10  Objective evaluation results of different image dehazing algorithms on real aerial fog images ((a) BIQME; (b) FADE; (c) VI; (d) RI; (e) CB; (f) VIF; (g) GB; (h) Entropy)

    图  11  消融实验结果 ((a) 航拍雾图像; (b) 本文方法; (c) 缺少L1L2; (d) 缺少LS; (e) 缺少LTV; (f) 无多尺度(Mutil-Scale, MS)结构; (g) 无MS结构和L (损失函数为均方误差); (h) AOD-Net)

    Fig.  11  Experimental results of ablation study ((a) Aerial fog image; (b) Proposed algorithm; (c) w/o L1 and L2 loss function; (d) w/o LS loss function; (e) w/o LTV loss function; (f) w/o MS structure; (g) w/o MS structure and L (The loss function is the mean square error); (h) AOD-Net)

    图  12  所提算法在λ2不同取值时的去雾效果

    Fig.  12  The dehazing effect of the proposed algorithm at different values of λ2

    图  13  $ {\lambda _2} $不同取值时的图像指标变化($ {\lambda _2} $=0.1 ~ 0.9)

    Fig.  13  The change of image index when λ2 takes different value ($ {\lambda _2} $=0.1 ~ 0.9)

    图  14  $ {\lambda _2} $不同取值时的图像指标变化($ {\lambda _2} $=0.81 ~ 0.89)

    Fig.  14  The change of image index when$ {\lambda _2} $takes different value ($ {\lambda _2} $=0.81 ~ 0.89)

    图  15  航拍去雾图像中车辆检测结果示例 ((a) 原图; (b) DCP; (c) BCCR; (d) CAP; (e) DehazeNet; (f) MSCNN;(g) AOD-Net; (h) GFN; (i) GCANet; (j) FFANet; (k) 本文算法)

    Fig.  15  Example of aerial image dehazing in vehicle detection ((a) Original; (b) DCP; (c) BCCR; (d) CAP; (e) DehazeNet; (f) MSCNN; (g) AOD-Net; (h) GFN; (i) GCANet; (j) FFANet; (k) Proposed algorithm)

    图  16  清晰图像上的去雾效果对比( (a) 原图; (b) DCP; (c) BCCR; (d) CAP; (e) DehazeNet; (f) MSCNN; (g) AOD-Net; (h) GFN; (i) GCANet; (j) FFANet; (k) 本文算法)

    Fig.  16  Comparison of dehazing effects on clear images ((a) Original; (b) DCP; (c) BCCR; (d) CAP; (e) DehazeNet; (f) MSCNN; (g) AOD-Net; (h) GFN; (i) GCANet; (j) FFANet; (k) Proposed algorithm)

    图  17  去雾失败时的示例图

    Fig.  17  The example images of failure in defogging

    表  1  本文所提网络的参数

    Table  1  The architectures of proposed network

    LayerInput SizeNumFilterPad
    Conv1128×128×3321×10
    Conv2128×128×32321×10
    Pool1128×128×323×31
    Concat1128×128×64
    Conv3128×128×64321×10
    Pool2128×128×325×52
    Concat2128×128×96
    Conv4128×128×96321×10
    Pool3128×128×327×73
    Concat3128×128×128
    Conv5128×128×12831×10
    下载: 导出CSV

    表  2  在合成有雾图像上的SSIM与PSNR结果

    Table  2  Comparison of SSIM and PSNR tested on synthetic fog images

    ModelSSIMPSNR (dB)
    IndoorOutdoorAverageIndoorOutdoorAverage
    DCP[6]0.74180.79010.765917.314815.532316.4236
    BCCR[7]0.80880.77190.790417.211916.304116.7580
    CAP[8]0.79420.82550.809916.949619.082918.0163
    DehazeNet[11]0.86530.83170.848520.054521.999221.0269
    MSCNN[12]0.77960.79310.786416.892719.001917.9473
    AOD-Net[13]0.83340.88480.859120.026718.410519.2186
    GFN[14]0.90390.88140.892721.107925.539923.3239
    GCANet[15]0.90250.85820.880422.336326.143124.2397
    FFANet[16]0.93130.90820.919828.405727.993228.1995
    Proposed0.87940.90110.890321.192423.407322.8994
    下载: 导出CSV

    表  3  在真实航拍雾图上的客观数值评价: (1) 最好的结果; (2) 次好的结果; (3) 第三好的结果

    Table  3  Objective numerical evaluation on the fog map of real aerial photography: (1) The best result; (2) The second-best result; (3) The third-best result

    ModelBIQMEFADEVIRICBVIFGBEntropy
    DCP[6]0.62350.69020.8992 (1)0.94070.38840.99570.58147.4427
    BCCR[7]0.57510.70910.86390.92630.42390.92110.51127.3983
    CAP[8]0.63800.82060.80260.92010.33980.88920.64137.4115
    DehazeNet[11]0.6917 (3)0.59840.8845 (2)0.9496 (2)0.42031.03960.64487.7265
    MSCNN[12]0.7131 (2)0.62610.8721 (3)0.9425 (3)0.44391.1513 (1)0.73527.3524
    AOD-Net[13]0.60130.84590.83650.92080.39950.95320.69687.4785
    GFN[14]0.7795 (1)0.4119 (1)0.85730.9498 (1)0.4880 (1)1.08720.8065 (2)7.8334 (3)
    GCANet[15]0.62440.4743 (2)0.85120.93280.4566 (3)1.1473 (2)0.75187.8891 (1)
    FFANet[16]0.63970.97520.81150.90740.44291.07240.9127 (1)7.3193
    Proposed0.66940.5625 (3)0.86490.94110.4753 (2)1.1109 (3)0.7749 (3)7.8799 (2)
    下载: 导出CSV

    表  4  消融实验中的数值指标

    Table  4  The numerical index in ablation experiment

    Modelw/o partSSIMPSNRVIRIFADEGB
    ProposedMS0.734426.64190.81950.90040.69710.6806
    L1 and L20.765928.04170.85470.91980.58320.7612
    LS0.717727.12590.80260.90350.70430.6327
    LTV0.739221.78840.69520.72590.88260.6122
    MS and L0.729325.53320.79190.88270.70330.6303
    0.890328.89940.89960.93280.51080.7749
    AOD-Net0.703123.39030.75230.90020.74390.7114
    下载: 导出CSV

    表  5  去雾耗时对比(s)

    Table  5  Comparison of the defogging time-cost (s)

    ModelAerial image size
    640×4801280×720
    DCP[6]5.995415.3855
    BCCR[7]1.77315.1932
    CAP[8]1.19214.4937
    DehazeNet[11]1.81335.5399
    MSCNN[12]1.59618.1504
    AOD-Net[13]0.00570.0423
    GFN[14]0.03790.2158
    GCANet[15]0.09530.4964
    FFANet[16]0.39322.7613
    Proposed0.02790.2084
    下载: 导出CSV

    表  6  参数量与模型大小比较

    Table  6  Comparison of the parameters and model size

    ModelParaSizePlatform
    DCP[6]Matlab
    BCCR[7]Matlab
    CAP[8]Matlab
    DehazeNet[11]8240Matlab
    MSCNN[12]8014Matlab
    AOD-Net[13]18338.9 KBPytorch
    GFN[14]5144151.99 MBCaffe
    GCANet[15]165227384.6 MBPytorch
    FFANet[16]445591357.85 MBPytorch
    Proposed21813286.5 KBCaffe
    下载: 导出CSV

    表  7  车辆检测的置信度数值, 标出三个最优的结果: (1) 最好的结果; (2) 次好的结果; (3) 第三好的结果

    Table  7  Confidence value of vehicle detection, marking the three best results: (1) The best result; (2) The second-best result; (3) The third-best result

    Dehazing modelDetection confidence before dehazingDetection confidence after dehazingThe number of identified carDetection confidence improvement
    Original0.68373.74
    DCP[6]0.79175.4215.80%
    BCCR[7]0.8233 (2)7.84 (1)20.42% (2)
    CAP[8]0.64676.26−5.41%
    DehazeNet[11]0.8033 (3)7.32 (3)17.49% (3)
    MSCNN[12]0.78354.3914.60%
    AOD-Net[13]0.64715.53−5.35%
    GFN[14]0.72275.955.70%
    GCANet[15]0.79804.2516.72%
    FFANet[16]0.75395.5010.27%
    Proposed0.8362 (1)7.48 (2)22.31% (1)
    下载: 导出CSV

    表  8  清晰航拍图像上的客观数值评价

    Table  8  Objective numerical evaluation on the clear images of aerial photography

    ModelClear images
    PSNRSSIM
    DCP[6]23.37310.8194
    BCCR[7]18.32050.7397
    CAP[8]21.42640.7468
    DehazeNet[11]23.23280.8664
    MSCNN[12]26.11350.8829
    AOD-Net[13]22.43630.8405
    GFN[14]26.95930.8869
    GCANet[15]26.40190.8497
    FFANet[16]29.33780.9299
    Proposed28.05440.9091
    下载: 导出CSV
  • [1] 韩敏, 闫阔, 秦国帅. 基于改进KAZE的无人机航拍图像拼接算法. 自动化学报, 2019, 45(2): 305-314

    . Han Ming, Yan Kuo, Qin Guo-Shuai. A mosaic algorithm for UAV aerial image with improved KAZE. Acta Automatica Sinica, 2019, 45(2): 305-314
    [2] Patsiouras E, Tefas A, Pitas I. Few-shot image recognition for UAV sports cinematography. In: Proceedings of 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW). Seattle, WA, USA: IEEE, 2020. 965−969
    [3] 周剑, 贾金岩, 张震, 陈盛伟. 面向应急保障的 5G 网联无人机关键技术. 重庆邮电大学学报(自然科学版), 2020, 32(4): 511-518

    . Zhou Jian, Jia Jin-Yan, Zhang Zhen, Chen Sheng-Wei. Key technologies for emergency communication based on 5G networked UAVs. Journal of Chongqing University of Posts and Telecommunications (Natural Science Edition), 2020, 32(4): 511-518
    [4] Liu P J, Horng S J, Lin J S, Li T R. Contrast in haze removal: configurable contrast enhancement model based on dark channel prior. IEEE Transactions on Image Processing, 2019, 28(5): 2212−2227
    [5] . Bui M T, Kim W. Single image dehazing using color ellipsoid prior. IEEE Transactions on Image Processing, 2018, 27(2): 999-1009 doi: 10.1109/TIP.2017.2771158
    [6] . He K M, Sun J, Tang X O. Single image haze removal using dark channel prior. IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), 2011, 33(12): 2341-2353 doi: 10.1109/TPAMI.2010.168
    [7] Meng G F, Wang Y, Duan J Y, Xiang S M, Pan C H. Efficient image dehazing with boundary constraint and contextual regularization. In: Proceedings of the 2013 IEEE International Conference on Computer Vision (ICCV). Sydney, Australia: IEEE, 2013. 617−624
    [8] . Zhu Q S, Mai J M, Shao L. A fast single image haze removal algorithm using color attenuation prior. IEEE Transactions on Image Processing, 2015, 24(11): 3522-3533 doi: 10.1109/TIP.2015.2446191
    [9] Chen W, Chen R Q, Lu Y, Yan Y, Wang H Z. Recurrent context aggregation network for single image dehazing. IEEE Transactions on Signal Processing Letters, 2021, 28: 419−423
    [10] . Chen W T, Fang H Y, Ding J J, Kuo S Y. PMHLD: Patch map-based hybrid learning DehazeNet for single image haze removal. IEEE Transactions on Image Processing, 2020, 29: 6773-6788 doi: 10.1109/TIP.2020.2993407
    [11] . Cai B L, Xu X M, Jia K, Qing C M, Tao D C. DehazeNet: an end-to-end system for single image haze removal. IEEE Transactions on Image Processing, 2016, 25(11): 5187-5198 doi: 10.1109/TIP.2016.2598681
    [12] Ren W Q, Liu S, Zhang H, Pan J S. Single image dehazing via multi-scale convolutional neural networks. In: Proceedings of the 2016 European Conference on Computer Vision (ECCV). Amsterdam, Netherlands: Springer, 2016. 154−169
    [13] Li B Y, Peng X L, Wang Z Y, Xu J Z, Feng D. AOD-Net: all-in-one dehazing network. In: Proceedings of the 2017 IEEE International Conference on Computer Vision (ICCV). Venice, Italy: IEEE, 2017. 4780−4788
    [14] Ren W Q, Ma L, Zhang J W, Pan J S, Cao X C, Liu W, et al. Gated fusion network for single image dehazing. In: Proceedings of the 2018 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Salt Lake City, UT, USA: IEEE, 2018. 3253−3261
    [15] Chen D D, He M M, Fan Q N, Liao J, Zhang L H, Hou D D, et al. Gated context aggregation network for image dehazing and deraining. In: Proceedings of the 2019 IEEE Winter Conference on Applications of Computer Vision (WACV). Waikoloa, HI, USA: IEEE, 2019. 1375−1383
    [16] Qin X, Wang Z L, Bai Y C, Xie X D, Jia H Z. FFA-Net: Feature fusion attention network for single image dehazing. In: Proceedings of the Association for the Advance of Artificial Intelligence. Hilton Midtown, New York: AAAI, 2020. 11908−11915
    [17] 杨燕, 陈高科, 周杰. 基于高斯权重衰减的迭代优化去雾算法. 自动化学报, 2019, 45(4): 819-828

    . Yang Yan, Chen Gao-Ke, Zhou Jie. Iterative optimization defogging algorithm using gaussian weight decay. Acta Automatica Sinica, 2019, 45(4): 819-828
    [18] Zhang J, Cao Y, Wang Y, Wen C L, Chen W C. Fully point-wise convolutional neural network for modeling statistical regularities in natural images. In: Proceedings of the 26th ACM International Conference on Multimedia. Seoul, Korea: ACM, 2018. 984−992
    [19] Silberman N, Hoiem D, Kohli P, Fergus R. Indoor segmentation and support inference from RGBD images. In: Proceedings of the 12th European conference on Computer Vision-Volume Part V. Berlin, Heidelberg: Springer, 2012. 746−760
    [20] . Ephraim Y, Malah D. Speech enhancement using a minimum mean-square error log-spectral amplitude estimator. IEEE Transactions on Acoustics, Speech, and Signal Processing, 1985, 33(2): 443-445 doi: 10.1109/TASSP.1985.1164550
    [21] . Zhang J, Tao D C. FAMED-Net: A fast and accurate multi-scale end-to-end dehazing network. IEEE Transactions on Image Processing, 2020, 29: 72-84 doi: 10.1109/TIP.2019.2922837
    [22] Guo C L, Li C Y, Guo J C, Loy C, Hou J H, Kwong S, et al. Zero-reference deep curve estimation for low-light image enhancement. In: Proceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Seattle, WA, USA: IEEE, 2020. 1777−1786
    [23] Zhang H, Patel V M. Densely connected pyramid dehazing network. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Salt Lake City, UT, USA: IEEE, 2018. 3194−3203
    [24] Das S, Dutta S. Fast deep multi-patch hierarchical network for nonhomogeneous image dehazing. In: Proceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW). Seattle, WA, USA: IEEE, 2020. 1994−2001
    [25] Chen S X, Chen Y Z, Qu Y Y, Huang J Y, Hong M. Multi-scale adaptive dehazing network, In: Proceedings of the 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW). Long Beach, CA, USA: IEEE, 2019. 2051−2059
    [26] . Zhou W, Bovik A, Sheikh H, Simoncelli E. Image quality assessment: from error visibility to structural similarity. IEEE Transactions on Image Processing, 2004, 13(4): 600-612 doi: 10.1109/TIP.2003.819861
    [27] Justin J, Alexandre A, Li F F. Perceptual losses for real-time style transfer and super-resolution. In: Proceedings of the 2016 European conference on computer vision (ECCV). Amsterdam, Netherlands: Springer, 2016. 694−711
    [28] . Zhao H, Gallo O, Frosio I, Kautz J. Loss functions for image restoration with neural networks. IEEE Transactions on Computational Imaging, 2017, 3(1): 47-57 doi: 10.1109/TCI.2016.2644865
    [29] . Li B Y, Ren W Q, Fu D P, Tao D C, Feng D, Zeng W J, et al. Benchmarking single-image dehazing and beyond. IEEE Transactions on Image Processing, 2019, 28(1): 492-505 doi: 10.1109/TIP.2018.2867951
    [30] Gu K, Tao D C, Qiao J F, Lin W S. Learning a no-reference quality assessment model of enhanced images with big data. IEEE Transactions on Neural Networks and Learning Systems, 2018, 29(4): 1301−1313
    [31] Choi L K, You J, Bovik A. Referenceless prediction of perceptual fog density and perceptual image defogging. IEEE Transations on Image Processing, 2015, 24(11): 3888−3901
    [32] Zhao S Y, Zhang L, Huang S Y, Shen Y, Zhao S J. Dehazing evaluation: Real-world benchmark datasets, criteria, and baselines. IEEE Transactions on Image Processing, 2020, 29, 6947−6962
    [33] . Zhu Z Q, Wei H Y, Hu Gang, Li Y Y, Qi G Q, et al. A novel fast single image dehazing algorithm based on artificial multiexposure image fusion. IEEE Transactions on Instrumentation and Measurement. 2021, 70, 5001523
    [34] Liu Z, Blasch E, Xue Z Y, Zhao J Y, Laganiere R, et al. Objective assessment of multiresolution image fusion algorithms for context enhancement in night vision: A comparative study. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2012, 34(1): 94−109
    [35] Sheikh H, Bovik A. Image information and visual quality. IEEE Transactions on Image Processing, 2006, 15(2): 430−444
    [36] Xydeas C, Petrovic V. Objective image fusion performance measure. Electronics Letters, 2000, 36(4): 308−309
    [37] 程宇, 邓德祥, 颜佳, 范赐恩. 基于卷积神经网络的弱光照图像增强算法. 计算机应用, 2019, 39(4): 1162-1169 doi: 10.11772/j.issn.1001-9081.2018091979

    . Chen Yu, Deng De-Xiang, Yan Jia, Fan Ci-En. Weakly illuminated image enhancement algorithm based on convolutional neural network. Journal of Computer Applications, 2019, 39(4): 1162-1169 doi: 10.11772/j.issn.1001-9081.2018091979
    [38] Redmon J, Divvala S, Girshick R, Farhadi A. You only look once: unified, real-time object detection. In: Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Las Vegas, NV, USA: IEEE, 2016. 779−788
  • 加载中
图(17) / 表(8)
计量
  • 文章访问数:  1741
  • HTML全文浏览量:  831
  • PDF下载量:  378
  • 被引次数: 0
出版历程
  • 收稿日期:  2021-03-24
  • 录用日期:  2021-12-02
  • 网络出版日期:  2021-12-12
  • 刊出日期:  2022-06-02

目录

    /

    返回文章
    返回