2.624

2020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

基于差异激励的无参考图像质量评价

陈勇 吴明明 房昊 刘焕淋

陈勇, 吴明明, 房昊, 刘焕淋. 基于差异激励的无参考图像质量评价. 自动化学报, 2020, 46(8): 1727−1737 doi: 10.16383/j.aas.c180088
引用本文: 陈勇, 吴明明, 房昊, 刘焕淋. 基于差异激励的无参考图像质量评价. 自动化学报, 2020, 46(8): 1727−1737 doi: 10.16383/j.aas.c180088
Chen Yong, Wu Ming-Ming, Fang Hao, Liu Huan-Lin. No-reference image quality assessment based on differential excitation. Acta Automatica Sinica, 2020, 46(8): 1727−1737 doi: 10.16383/j.aas.c180088
Citation: Chen Yong, Wu Ming-Ming, Fang Hao, Liu Huan-Lin. No-reference image quality assessment based on differential excitation. Acta Automatica Sinica, 2020, 46(8): 1727−1737 doi: 10.16383/j.aas.c180088

基于差异激励的无参考图像质量评价

doi: 10.16383/j.aas.c180088
基金项目: 

国家自然科学基金 60975008

重庆市研究生科研创新项目 CYS17235

详细信息
    作者简介:

    陈勇  重庆邮电大学自动化学院教授. 2003年获得重庆大学机械工程专业博士.主要研究方向为图像处理与模式识别, 智能优化控制. E-mail: chenyong@cqupt.edu.cn

    吴明明  重庆邮电大学自动化学院硕士研究生.主要研究方向为图像处理, 机器学习与计算机视觉. E-mail:darcy.wu@unisoc.com

    房昊   重庆邮电大学自动化学院硕士研究生.主要研究方向为图像处理, 机器学习与计算机视觉. E-mail: f1010507348@163.com

    通讯作者:

    刘焕淋  重庆邮电大学通信与信息工程学院教授. 2008年获得重庆大学光电工程专业博士.主要研究方向为信息获取与处理, 机器学习与计算机视觉.本文通信作者. E-mail: liuhl@cqupt.edu.cn

No-reference Image Quality Assessment Based on Differential Excitation

Funds: 

National Natural Science Foundation of China 60975008

Chongqing Graduate Student Science Research Innovation of China CYS17235

More Information
    Author Bio:

    CHEN Yong    Professor at the School of Automation, Chongqing University of Posts and Telecommunication. He received his Ph. D. degree in mechanical engineering from Chongqing University in 2003. His research interest covers image processing, pattern recognition, and intelligent optimizing controls

    WU Ming-Ming   Master student at the School of Automation, Chongqing University of Posts and Telecommunication. His research interest covers image processing, machine learning, and computer vision

    FANG Hao    Master student at the School of Automation, Chongqing University of Posts and Telecommunication. His research interest covers image processing, machine learning, and computer vision

    Corresponding author: LIU Huan-Lin  Professor at the School of Telecommunication and Information Engineering, Chongqing University of Posts and Telecommunication. She received her Ph. D. degree in optical engineering from Chongqing University in 2008. Her research interest covers information acquiring and processing, machine learning, and computer vision. Corresponding author of this paper
  • 摘要: 为了衡量图像的降质程度, 充分考虑像素间的相关性, 提出了一种基于差异激励的无参考图像质量评价算法.该算法根据韦伯定律求得差异激励图, 并依据各向异性得到差异激励的梯度映射图; 然后量化差异激励得到差异量化图, 并分别与差异激励图与梯度映射图进行加权融合; 最后利用求得的特征, 通过支持向量回归(Support vector regression, SVR)预测得出图像质量的客观评价值.在LIVE、MLIVE、MDID2013和MDID2016等多个数据库中测试显示, 该算法稳定性强, 复杂度低, 能准确反映人类对图像质量的视觉感知效果.
    Recommended by Associate Editor LIU Yue-Hu
    1)  本文责任编委 刘跃虎
  • 图  1  不同程度失真图像对应的差异激励图和梯形映射图

    Fig.  1  Difference excitation diagram and trapezoid map corresponding to different degree distortion images

    图  2  定向的梯度滤波器

    Fig.  2  Directional gradient filter

    图  3  差异量化图与局部量化值

    Fig.  3  Difference quantization map and local quantization value

    图  4  差异量化图的概率统计直方图

    Fig.  4  Probability statistical histogram of difference quantization map

    图  5  本文算法流程框图

    Fig.  5  The algorithm flow chart in this paper

    表  1  本文选用的6个图像库描述

    Table  1  The descriptions of six image databases selected in this paper

    图像库 参考图像 失真类型 图像个数
    LIVE 29 JPEG2000压缩 953
    JPEG压缩
    高斯白噪声
    高斯模糊
    快衰弱
    CSIQ 30 加性高斯噪声 900
    高斯模糊
    对比度改变
    粉红噪声
    MLIVE 15 JPEG压缩 450
    JPEG2000压缩
    模糊+压缩
    模糊+噪声
    MDID2013 12 模糊+压缩+噪声 324
    MDID2016 20 噪声+模糊+对比度+压缩+ JP2K压缩 1 600
    TID2013 25 #1加性高斯噪声 #13 JPEG2000传输误差 3 000
    #2彩色分量中的差分加性噪声 #14无偏心率类型噪声
    #3空域相关噪声 #15不同强度局部块失真
    #4掩膜噪声 #16均值平移
    #5高频噪声 #17对比度改变
    #6脉冲噪声 #18色彩饱和度改变
    #7量化噪声 #19乘性高斯噪声
    #8高斯模糊 #20舒适噪声
    #9图像去噪 #21噪声图像的有损压缩
    #10 JPEG压缩 #22图像的颜色量化及波动
    #11 JPEG2000压缩 #23图像色差
    #12 JPEG传输误差 #24稀疏采样及重构
    下载: 导出CSV

    表  2  LIVE和CSIQ数据库中单一型算法质量评价性能对比

    Table  2  Comparison of performance evaluation of single algorithm in LIVE and CSIQ databases

    LIVE (953 images)CSIQ (900 images)
    算法CCSROCCRMSECCSROCCRMSE
    DIIVINE[5]0.8930.88511.1680.7970.8100.275
    BRISQUE[6]0.9440.9477.7950.7280.7400.325
    NIQE[7]0.9090.90811.3760.7560.7390.340
    IL-NIQE[8]0.9060.90310.8240.7320.7180.354
    NR-GLBP[9]0.9420.9359.0750.8470.8010.174
    NRSL[10]0.9570.9538.0180.8590.8510.109
    本文算法0.9630.9617.0520.8580.8390.117
    下载: 导出CSV

    表  3  TID2013数据库中算法质量评价性能指标SROCC对比(3 000幅图)

    Table  3  Comparison of quality evaluation performance indexes of algorithm in TID2013 database (3 000 images)

    算法 #1 #2 #3 #4 #5 #6 #7 #8 #9 #10 #11 #12 #13 #14 #15 #16 #17 #18 #19 #20 #21 #22 #23 #24 All
    BRISQUE[6] 0.706 0.523 0.776 0.295 0.836 0.802 0.682 0.861 0.500 0.790 0.779 0.254 0.723 0.213 0.197 0.217 0.079 0.113 0.674 0.198 0.627 0.849 0.724 0.811 0.567
    NR-GLBP[9] 0.466 0.591 0.759 0.491 0.875 0.693 0.833 0.878 0.721 0.844 0.867 0.440 0.594 0.226 0.204 0.105 0.123 0.023 0.580 0.447 0.507 0.762 0.748 0.830 0.679
    NRSL[10] 0.813 0.457 0.867 0.393 0.902 0.787 0.700 0.886 0.795 0.818 0.891 0.345 0.805 0.117 0.323 0.136 0.194 0.110 0.753 0.434 0.751 0.866 0.694 0.887 0.661
    NFERM[13] 0.851 0.520 0.846 0.521 0.894 0.857 0.785 0.888 0.741 0.797 0.920 0.381 0.718 0.176 0.081 0.238 0.056 0.029 0.762 0.206 0.401 0.848 0.684 0.878 0.652
    GWH-GLBP[14] 0.736 0.358 0.814 0.412 0.874 0.795 0.757 0.838 0.811 0.890 0.901 0.494 0.656 0.326 0.344 0.341 0.252 0.420 0.601 0.624 0.664 0.741 0.919 0.898 0.655
    本文算法 0.768 0.454 0.861 0.537 0.885 0.814 0.752 0.908 0.859 0.853 0.940 0.544 0.754 0.426 0.480 0.275 0.442 0.507 0.706 0.680 0.823 0.839 0.948 0.903 0.691
    下载: 导出CSV

    表  4  MLIVE、MDID2013和MDID2016数据库中混合型算法质量评价性能指标对比

    Table  4  Comparison of quality evaluation performance indicators of hybrid algorithm in MLIVE, MDID2013 and MDID2016 databases

    算法MLIVEMDID2013MDID2016
    (450 images)(324 images)(1 600 images)
    CCSROCCRMSECCSROCCRMSECCSROCCRMSE
    SISBLM[10]0.9250.9077.1980.9100.9050.0190.6330.6551.708
    HOSA[12]0.9260.9026.9740.8920.8720.0210.5660.5511.871
    NFERM[13]0.9190.8997.4580.8710.8550.0250.4960.4511.915
    GWH-GLBP[14]0.9450.9396.0610.9130.9070.0190.8910.8861.004
    本文算法0.9570.9425.7360.9160.9040.0190.9030.8920.947
    下载: 导出CSV

    表  5  测试不同训练与测试比例的SROCC和CC的中值(1 000次)

    Table  5  Median values of SROCC and CC for different training and test ratios (1 000 times)

    LIVEMLIVE
    测试集和训练集比例指标JP2KJPEGGBLURFFWNALLGB + JPEGGB + WNALL
    70 $\%$与30 $\%$SROCC0.94960.95920.94190.88480.97520.96040.93720.94380.9366
    CC0.96130.97680.95040.89580.98040.96240.95980.95350.9506
    60 $\%$与40 $\%$SROCC0.94790.95700.93970.87320.97470.95780.92660.93210.9278
    CC0.95940.97500.94680.88750.97960.96100.94830.94010.9414
    50 $\%$与50 $\%$SROCC0.94660.95460.93630.86740.97430.95590.92040.92400.9223
    CC0.95870.97380.94130.88390.97910.95950.94370.93570.9364
    40 $\%$与60 $\%$SROCC0.94090.95210.93200.86060.97060.95220.90630.90530.9046
    CC0.95200.96910.93690.87440.97890.95680.92430.91670.9164
    30 $\%$与70 $\%$SROCC0.93510.94650.92530.85140.96450.94630.90400.89840.8976
    CC0.94570.96510.92870.86110.97810.95070.92060.91010.9112
    20 $\%$与80 $\%$SROCC0.92530.93450.91380.83290.96120.93450.88640.87850.8811
    CC0.93510.95400.91500.84680.97480.93870.90790.88620.8918
    下载: 导出CSV

    表  6  CSIQ数据库中不同失真类型的性能评价

    Table  6  Performance evaluation of different distortion types in CSIQ database

    类型CCSROCCRMSEKROCC
    JP2K0.90460.87417.60440.6966
    JPEG0.93600.91946.50400.8151
    GBLUR0.88580.90167.84020.7320
    WN0.93770.92386.34930.7591
    ALL0.91670.89537.26870.7378
    下载: 导出CSV

    表  7  LIVE数据库中失真类型的识别准确率(1 000次)

    Table  7  Recognition accuracy of distortion type in LIVE database (1 000 times)

    类型JP2KJPEGGBLURFFWNALL
    准确率87.94 $\%$100 $\%$97.82 $\%$90.26 $\%$100 $\%$95.39 $\%$
    下载: 导出CSV

    表  8  图像质量评价算法运行时间

    Table  8  Running time of image quality evaluation algorithm

    IQA modelDIIVINEBRISQUENIQASISBLMHOSANFERMGWH-GLBP本文算法
    Time (s)0.1815.82.723.730.3555.10.270.33
    下载: 导出CSV
  • [1] 王朝云, 蒋刚毅, 郁梅, 陈芬.基于流形特征相似度的感知图像质量评价.自动化学报, 2016, 42(7): 1113-1124 doi: 10.16383/j.aas.2016.c150559

    Wang Chao-Yun, Jiang Gang-Yi, Yu Mei, Chen Fen. Manifold feature similarity based perceptual image quality assessment. Acta Automatica Sinica, 2016, 42(7): 1113-1124 doi: 10.16383/j.aas.2016.c150559
    [2] 南栋, 毕笃彦, 马时平, 凡遵林, 何林远.基于分类学习的去雾后图像质量评价算法.自动化学报, 2016, 42(2): 270-278 doi: 10.16383/j.aas.2016.c140854

    Nan Dong, Bi Du-Yan, Ma Shi-Ping, Fan Zun-Lin, He Lin-Yuan. A quality assessment method with classified-learning for Dehazed images. Acta Automatica Sinica, 2016, 42(2): 270-278 doi: 10.16383/j.aas.2016.c140854
    [3] 王志明.无参考图像质量评价综述.自动化学报, 2015, 41(6): 1062-1079 doi: 10.16383/j.aas.2015.c140404

    Wang Zhi-Ming. Review of no-reference image quality assessment. Acta Automatica Sinica, 2015, 41(6): 1062-1079 doi: 10.16383/j.aas.2015.c140404
    [4] 陈勇, 帅锋, 樊强.基于自然统计特征分布的无参考图像质量评价.电子信息学报, 2016, 38(7): 1645-1653 http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=dzkxxk201607010

    Chen Yong, Shuai Feng, Fan Qiang. A no-reference image quality assessment based on distribution characteristics of natural statistics. Journal of Electronics and Information Technology, 2016, 38(7): 1645-1653 http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=dzkxxk201607010
    [5] Moorthy A K, Bovik A C. Blind image quality assessment: From natural scene statistics to perceptual quality. IEEE Transactions on Image Processing, 2011, 20(12): 3350-3364 doi: 10.1109/TIP.2011.2147325
    [6] Mittal A, Moorthy A K, Bovik A C. No-reference image quality assessment in the spatial domain. IEEE Transactions on Image Processing, 2012, 21(12): 4695-4708 doi: 10.1109/TIP.2012.2214050
    [7] Mittal A, Soundararajan R, Bovik A C. Making a "completely blind" image quality analyzer. IEEE Signal Processing Letters, 2013, 20(3): 209-212 https://www.researchgate.net/publication/260636803_Making_a_Completely_Blind_Image_Quality_Analyzer
    [8] Zhang L, Zhang L, Bovik A C. A feature-enriched completely blind image quality evaluator. IEEE Transactions on Image Processing, 2015, 24(8): 2579-2591 doi: 10.1109/TIP.2015.2426416
    [9] Zhang M, Muramatsu C, Zhou X R, Hara T, Fujita H. Blind image quality assessment using the joint statistics of generalized local binary pattern. IEEE Signal Processing Letters, 2015, 22(2): 207-210 http://www.researchgate.net/publication/265555808_Blind_Image_Quality_Assessment_using_the_Joint_Statistics_of_Generalized_Local_Binary_Pattern
    [10] Li Q L, Lin W S, Xu J T, Fang Y M. Blind image quality assessment using statistical structural and luminance features. IEEE Transactions on Multimedia, 2016, 18(12): 2457-2469 doi: 10.1109/TMM.2016.2601028
    [11] 张敏辉, 杨剑.评价SAR图像去噪效果的无参考图像质量指标.重庆邮电大学学报(自然科学版), 2018, 30(04): 530-536 http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=cqydxyxb-zrkx201804014

    Zhang Min-Hui, Yang Jian. A new referenceless image quality index to evaluate denoising performance of SAR images. Journal of Chongqing University of Posts and Telecommunications: Natural Science Edition, 2018, 30(04): 530-536 http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=cqydxyxb-zrkx201804014
    [12] Xu J T, Ye P, Li Q H, Du H Q, Liu Y, Doermann D. Blind image quality assessment based on high order statistics aggregation. IEEE Transactions on Image Processing, 2016, 25(9): 4444-4457 doi: 10.1109/TIP.2016.2585880
    [13] Gu K, Zhai G T, Yang X K, Zhang W J. Using free energy principle for blind image quality assessment. IEEE Transactions on Multimedia, 2015, 17(1): 50-63 doi: 10.1109/TMM.2014.2373812
    [14] Li Q H, Lin W S, Fang Y M. No-reference quality assessment for multiply-distorted images in gradient domain. IEEE Signal Processing Letters, 2016, 23(4): 541-545
    [15] 卢彦飞, 张涛, 章程.应用log-Gabor韦伯特征的图像质量评价.光学精密工程, 2015, 23(11): 3259-3269 http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=gxjmgc201511031

    Lu Yan-Fei, Zhang Tao, Zhang Cheng. Image quality assessment using log-Gabor Weber feature. Optics and Precision Engineering, 2015, 23(11): 3259-3269 http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=gxjmgc201511031
    [16] 丁绪星, 朱日宏, 李建欣.一种基于人眼视觉特性的图像质量评价.中国图象图形学报, 2004, 9(2): 190-194 doi: 10.3969/j.issn.1006-8961.2004.02.012

    Ding Xu-Xing, Zhu Ri-Hong, Li Jian-Xin. A criterion of image quality assessment based on property of HVS. Journal of Image and Graphics, 2004, 9(2): 190-194 doi: 10.3969/j.issn.1006-8961.2004.02.012
    [17] Chen J, Shan S G, He C, Zhao G Y, Pietikainen M, Chen X L, et al. WLD: A robust local image descriptor. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2010, 32(9): 1705-1720 doi: 10.1109/TPAMI.2009.155
    [18] Sheikh H R, Wang Z, Cormack L, et al. LIVE image quality assessment database release 2[Online], available: http://ive.ece.utexas.edu/research/quality, October 22, 2005
    [19] Ding L, Huang H, Zang Y. Image quality assessment using directional anisotropy structure measurement. IEEE Transactions on Image Processing, 2017, 26(4): 1799-1809 http://www.onacademic.com/detail/journal_1000039816644010_5e60.html
    [20] Larsson J, Landy M S, Heeger D J. Orientation-selective adaptation to first- and second-order patterns in human visual cortex. Journal of Neurophysiology, 2006, 95(2): 862-881
    [21] Gu K, Zhai G T, Yang X K, Zhang W J. Hybrid no-reference quality metric for singly and multiply distorted images. IEEE Transactions on Broadcasting, 2014, 60(3): 555-567 doi: 10.1109/TBC.2014.2344471
    [22] Sun W, Zhou F, Liao Q M. MDID: A multiply distorted image database for image quality assessment. Pattern Recognition, 2017, 61(1): 153-168
    [23] Yu X, Yang E H, Wang H Q. Down-sampling design in DCT domain with arbitrary ratio for image/video transcoding. IEEE Transactions on Image Processing, 2009, 18(1): 75-89 http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=f1f3f5442589f45bf9a7d645a1fa5916
    [24] Wu F, Yu E, Yu P, Zhang K, Song Z. Modeling and prediction of the air permeability of fabrics based on the support vector machine. Journal of Testing and Evaluation, 2017, 45(4): 1388-1395 http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=48b8c07c52998562469b78f3b28312a5
    [25] Chang C C, Lin C J. LIBSVM: A library for support vector machines. ACM Transactions on Intelligent Systems and Technology (TIST), 2011, 2(3): Article No. 27
    [26] Ponomarenko N, Ieremeiev O, Lukin V, Egiazarian K, et al. Color image database TID2013: Peculiarities and preliminary results. In: Proceedings of the 4th Europian Workshop on Visual Information Processing EUVIP2013. Paris, France: 2013. 6
    [27] Mittal A, Soundararajan R, Bovik A C. Making a "om-pletely blind" image quality analyzer. IEEE Signal Process-ing Letters, 2013, 20(3): 209-212 http://www.researchgate.net/publication/260636803_Making_a_Completely_Blind_Image_Quality_Analyzer
  • 加载中
图(5) / 表(8)
计量
  • 文章访问数:  961
  • HTML全文浏览量:  140
  • PDF下载量:  127
  • 被引次数: 0
出版历程
  • 收稿日期:  2018-02-06
  • 录用日期:  2018-08-28
  • 刊出日期:  2020-08-26

目录

    /

    返回文章
    返回