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

评价彩色图像自动聚焦清晰度的互补色小波测度

周汶 李旦 张建秋

周汶, 李旦, 张建秋. 评价彩色图像自动聚焦清晰度的互补色小波测度. 自动化学报, 2020, 46(8): 1615−1627 doi: 10.16383/j.aas.c190513
引用本文: 周汶, 李旦, 张建秋. 评价彩色图像自动聚焦清晰度的互补色小波测度. 自动化学报, 2020, 46(8): 1615−1627 doi: 10.16383/j.aas.c190513
Zhou Wen, Li Dan, Zhang Jian-Qiu. A complementary color wavelet-based measure on color image sharpness assessment for autofocus. Acta Automatica Sinica, 2020, 46(8): 1615−1627 doi: 10.16383/j.aas.c190513
Citation: Zhou Wen, Li Dan, Zhang Jian-Qiu. A complementary color wavelet-based measure on color image sharpness assessment for autofocus. Acta Automatica Sinica, 2020, 46(8): 1615−1627 doi: 10.16383/j.aas.c190513

评价彩色图像自动聚焦清晰度的互补色小波测度

doi: 10.16383/j.aas.c190513
基金项目: 国家自然科学基金(11827808, 11974082), 上海市科技创新行动计划社会发展科技领域项目(19DZ1205805), 上海航天科技创新基金, 珠海复旦创新研究院项目资助
详细信息
    作者简介:

    周汶:复旦大学电子工程系硕士研究生. 主要研究方向为计算机视觉, 图像处理.E-mail: wenzhou17@fudan.edu.cn

    李旦:复旦大学电子工程系副教授. 主要研究方向为数字信号处理及应用. 本文通信作者.E-mail: lidan@fudan.edu.cn

    张建秋:复旦大学电子工程系教授. 主要研究方向为信号处理及其在通信、控制、测量、图像和雷达中的应用. E-mail: jqzhang@ieee.org

A Complementary Color Wavelet-based Measure on Color Image Sharpness Assessment for Autofocus

Funds: Supported by National Natural Science Foundation of China (11827808, 11974082), Social Development Project of Shanghai Science and Technology Innovation Action Plan (19DZ1205805), Shanghai Aerospace Science and Technology innovation Fund, Zhuhai-Fudan Innovation Research Institute Project
  • 摘要: 针对彩色图像的自动聚焦, 本文提出了一种新的清晰度评价测度. 该测度借助于互补色小波变换, 在互补色小波域, 利用融合互补色算子的层级最大能量和层级统计分布扩散度的乘积来描述本文的清晰度. 分析表明: 融合互补色算子, 可提取待评价彩色图像在颜色、亮度、方向、尺度和各通道分量间的相互信息等方面的相关特征. 这样, 其层级的最大能量就反映了这些被提取相关特征中最显著特征的清晰程度, 而其统计分布的扩散度, 就衡量了其清晰度特征分布的离散程度. 那么利用它们来共同表征清晰度的测度, 就使得本文所提的测度能随图像清晰程度的增加而增加. 在LIVE/IVC数据库上与多种经典方法的对比结果表明: 本文提出的测度具有最高的聚焦精度0.0373/0.0246、分辨率1.6132/0.4771和最好的无偏稳定性.
  • 图  1  HSI颜色空间与RGB色环

    Fig.  1  The HSI color space and the RGB hue ring

    图  6  层级1的方向子带的统计分布情况

    Fig.  6  Statistical distribution of directional subbands in level 1

    图  2  亮度值恒定的彩色图像

    Fig.  2  A color image with a constant intensity/brightness

    图  3  彩色图像在第2层级的(互补色)小波处理结果

    Fig.  3  Color image processing in level 2 using (Complementary color) wavelet transform

    图  4  MCCWT测度构建流程图($j = 1,2$表示互补色小波分解的层级; $n = k\pi /8,k = 1,2,\cdots,8$表示每一层级的8个子带的方向)

    Fig.  4  Modeling flow of MCCWT measure ($j = 1,2$ presents the decomposition level of the complementary color wavelet; $n = k\pi /8,k = 1,2,\cdots,8$ represents the subbands directions in each level)

    图  5  层级最大能量随模糊递增的变化趋势(实线为层级1, 虚线为层级2)

    Fig.  5  The change trend of the scale maximum energy with the increasing blur of color image (Solid line is level 1, dotted line is level 2)

    图  7  Woman hat图像1和图像3在第1层级、方向为$n = \pi /8$的扩散系数集合统计的真实分布及拟合分布(实线为真实分布, 虚线为拉普拉斯拟合分布)

    Fig.  7  Empirical and fitting distribution of diffusion coefficients of woman hat image 1 and image 3 with direction $n = \pi /8$ in level 1 (The solid line is the empirical distribution, and the dashed line is the Laplace fitting distribution)

    图  8  DIV2K图像序列层级分布的扩散度随模糊递增的变化趋势(实线为层级1, 虚线为层级2)

    Fig.  8  The change trend of the variability with the increasing blur of DIV2K image sequences (Solid line is level 1, dotted line is level 2)

    图  9  聚焦精度的定义

    Fig.  9  Definition of accuracy metrics

    图  10  MDB测度构建流程图($j = 1,2$表示小波分解层级; $n = k\pi /4,k = 1,2,3,4$表示每个层级子带的方向)

    Fig.  10  Modeling flow of MDB measure ($j = 1,2$ represents the decomposition level of the traditional wavelet; $n = k\pi /4,k = 1,2,3,4$ represents the subbands directions in each level)

    图  11  各清晰度评价方法在图像序列上的性能定性比较

    Fig.  11  Qualitative comparison of the performance of each sharpness assessment method in the Gaussian blur image sequences

    表  1  图像高斯模糊的方差及窗口大小

    Table  1  The variance and window size of image Gaussian blur

    图像$\sigma _h^2 $窗口大小图像$\sigma _h^2 $窗口大小
    10.283130.9647
    20.2931417
    30.33151.17
    40.323161.29
    50.333171.339
    60.343181.6711
    70.35319213
    80.43202.3315
    90.53212.6717
    100.66522319
    110.85233.3321
    120.9067243.6723
    下载: 导出CSV

    表  2  DIV2K中层级1、方向为$n = 4\pi /8$的子带的真实分布与拟合分布熵差

    Table  2  In DIV2K, the entropy difference between the empirical distribution and the fitting distribution of the subband with direction $n = 4\pi /8$ in level 1

    图像$\Delta H/H$图像$\Delta H/H$
    10.0254130.0278
    20.0253140.0281
    30.0252150.0290
    40.0246160.0303
    50.0243170.0225
    60.0239180.0367
    70.0235190.0415
    80.0224200.0681
    90.0241210.1853
    100.0259220.4329
    110.0265230.8159
    120.0274241.2271
    下载: 导出CSV

    表  3  各清晰度评价算法在LIVE数据库gblur图像序列中的聚焦精度和分辨率均值(e = 1%)

    Table  3  The average of accuracy metrics and resolution metrics of each sharpness assessment method in Gaussian blur image sequences of LIVE (gblur) database (e = 1%)

    Methods精度(AM)精度提升(%)分辨率(RM)分辨率提升(%)
    Entropy[3]5.462199.329.008182.09
    Tenengrad[6]0.113066.996.126273.67
    SMD[8]0.075650.666.105473.58
    SML[9]0.054331.313.274850.74
    Variance[15]0.579893.578.613381.27
    Sd-Svd[11]0.03801.842.132224.34
    DB[15]0.09500.746.515275.24
    FISH[14]0.063641.353.900458.64
    MDB0.04078.352.486235.11
    MCCWT0.03731.6132
    下载: 导出CSV

    表  4  各清晰度评价算法在IVC数据库Flou图像序列中的聚焦精度和分辨率均值(e = 1%)

    Table  4  The average of accuracy metrics and resolution metrics of each sharpness assessment method in Gaussian blur image sequences of IVC (Flou) database (e = 1%)

    Methods精度(AM)精度提升(%)分辨率(RM)分辨率提升(%)
    Entropy[3]2.534399.039.102794.47
    Tenengrad[6]0.078568.666.210592.32
    SMD[8]0.051752.426.064992.13
    SML[9]0.032323.841.721772.29
    Variance[15]0.406793.958.729194.53
    Sd-Svd[11]0.02657.172.263178.92
    DB[15]0.064561.866.492792.65
    FISH[14]0.040439.112.860183.32
    MDB0.029917.731.630170.73
    MCCWT0.02460.4771
    下载: 导出CSV

    表  5  各清晰度评价算法在LIVE数据库gblur图像序列中的最大值错误次数(噪声存在于图像2)

    Table  5  The number of errors in the maximum value of each sharpness assessment method in the Gaussian blur image sequence of LIVE (gblur) database (noise in image 2)

    Methods$ \sigma _n^2 = 5 $$ \sigma _n^2 = 10 $$ \sigma _n^2 = 15 $$ \sigma _n^2 = 20 $
    Entropy[3]13222427
    Tenengrad[6]072024
    SMD[8]8232728
    SML[9]8222628
    Variance[15]00613
    Sd-Svd[11]8232728
    DB[15]6202528
    FISH[14]2202528
    MDB0009
    MCCWT0000
    下载: 导出CSV

    表  6  各清晰度评价算法在LIVE数据库gblur图像序列中的最大值错误次数(噪声存在于图像3)

    Table  6  The number of errors in the maximum value of each sharpness assessment method in the Gaussian blur image sequence of LIVE (gblur) database (noise in image 3)

    Methods$ \sigma _n^2 = 5 $$ \sigma _n^2 = 10 $$ \sigma _n^2 = 15 $$ \sigma _n^2 = 20 $
    Entropy[3]13242627
    Tenengrad[6]0122125
    SMD[8]11232828
    SML[9]8232628
    Variance[15]01813
    Sd-Svd[11]8242828
    DB[15]8222628
    FISH[14]2222628
    MDB00314
    MCCWT0001
    下载: 导出CSV

    表  7  各清晰度评价算法在LIVE数据库gblur图像序列中的最大值错误次数(噪声存在于图像4)

    Table  7  The number of errors in the maximum value of each sharpness assessment method in the Gaussian blur image sequence of LIVE (gblur) database (noise in image 4)

    Methods$ \sigma _n^2 = 5 $$ \sigma _n^2 = 10 $$ \sigma _n^2 = 15 $$ \sigma _n^2 = 20 $
    Entropy[3]17252827
    Tenengrad[6]0132326
    SMD[8]11252828
    SML[9]10232828
    Variance[15]051115
    Sd-Svd[11]8242828
    DB[15]10222728
    FISH[14]6232828
    MDB00921
    MCCWT0002
    下载: 导出CSV

    表  8  各清晰度评价算法在LIVE数据库gblur图像序列中的平均运行时间

    Table  8  Average running time of each sharpness assessment method in LIVE (gblur) database

    MethodsTime (s)MethodsTime (s)
    Entropy[3]2.6349Sd-Svd[11]0.1865
    Tenengrad[6]0.0371DB[15]0.0831
    SMD[8]0.0283FISH[14]0.0806
    SML[9]0.0234MDB0.1417
    Variance[15]0.0084MCCWT0.7736
    下载: 导出CSV

    表  9  各清晰度评价算法在IVC数据库Flou图像序列中的平均运行时间

    Table  9  Average running time of each sharpness assessment method in IVC (Flou) database

    MethodsTime (s)MethodsTime (s)
    Entropy[3]2.1115Sd-Svd[11]0.1712
    Tenengrad[6]0.0294DB[15]0.0718
    SMD[8]0.0222FISH[14]0.0658
    SML[9]0.0203MDB0.1138
    Variance[15]0.0063MCCWT0.6111
    下载: 导出CSV
  • [1] 谢攀, 张利, 康宗明, 谢时根. 一种基于尺度变化的DCT自动聚焦算法. 清华大学学报(自然科学版), 2003, 43(1): 55−58 doi: 10.3321/j.issn:1000-0054.2003.01.017

    Xie Pan, Zhang Li, Kang Zong-Ming, Xie Shi-Gen. Window scale-based automatic focus algorithm using DCT transformation. Journal of Tsinghua University (Science and Technology), 2003, 43(1): 55−58 doi: 10.3321/j.issn:1000-0054.2003.01.017
    [2] 康宗明, 张利, 谢攀. 一种基于能量和熵的自动聚焦算法. 电子学报, 2003, 31(4): 552−555 doi: 10.3321/j.issn:0372-2112.2003.04.020

    Kang Zong-Ming, Zhang Li, Xie Pan. Implementation of an automatic focusing algorithm based on spatial high frequency energy and entropy. Acta Electronica Sinica, 2003, 31(4): 552−555 doi: 10.3321/j.issn:0372-2112.2003.04.020
    [3] 贾旭, 曹玉东, 孙福明, 崔建江, 薛定宇. 基于无参考质量评价模型的静脉图像采集方法. 电子学报, 2015, 43(2): 236−241 doi: 10.3969/j.issn.0372-2112.2015.02.005

    Jia Xu, Cao Yu-Dong, Sun Fu-Ming, Cui Jian-Jiang, Xue Ding-Yu. Vein image acquisition method based on quality assessment model without reference. Acta Electronica Sinica, 2015, 43(2): 236−241 doi: 10.3969/j.issn.0372-2112.2015.02.005
    [4] Firestone L, Cook K, Culp K, Talsania N, Preston K. Comparison of autofocus methods for automated microscopy. Cytometry, 1991, 12(3): 195−206 doi: 10.1002/cyto.990120302
    [5] Choi K S, Lee J S, Ko S J. New autofocusing technique using the frequency selective weighted median filter for video cameras. IEEE Transactions on Consumer Electronics, 1999, 45(3): 820−827 doi: 10.1109/30.793616
    [6] Tenenbaum J M. Accummodation in Computer Vision, [Ph. D. dissertation], Stanford University, USA, 1970
    [7] Zhan Y B, Zhang R. No-Reference image sharpness assessment based on maximum gradient and variability of Gradients. IEEE Transactions on Multimedia, 2018, 20(7): 1796−1808
    [8] Jarvis R A. Focus optimization criteria for computer image processing. Microscope, 1976, 24(2): 163−180
    [9] Nayar S K, Nakagawa Y. Shape from focus system. IEEE Tansactions Pattern Analysis and Machine Intelligence, 1994, 16(8): 824−831 doi: 10.1109/34.308479
    [10] 李郁峰, 陈念年, 张佳成. 一种快速高灵敏度聚焦评价函数. 计算机应用研究, 2010, 27(4): 1534−1536

    Li Yu-Feng, Chen Nian-Nian, Zhang Jia-Cheng. Fast and high sensitivity focusing evaluation function. Application Research of Computers, 2010, 27(4): 1534−1536
    [11] Li Y J, Di X G. A no-reference infrared image sharpness assessment based on singular value decomposition. In: Proceedings of the 2016 Visual Communication and Image Processing. Chengdu China: IEEE, 2016. 1−4
    [12] Baina J, Dublet J. Automatic focus and iris control for video cameras. In Proceedings of the Fifth International Conference Image Processing and its Applications. Edinburgh, UK: IEEE, 1995, 232−235
    [13] 菅维乐, 姜威, 周贤. 一种基于小波变换的数字图像自动聚焦算法. 山东大学学报: 工学版, 2004, 34(6): 38−40

    Jian Wei-Le, Jiang Wei, Zhou Xian. Auto-focusing algorithms of aigital image based on wavelet transform. Journal of Shandong University (Engineering Science), 2004, 34(6): 38−40
    [14] Vu P V, Chandler D M. A fast wavelet-based algorithm for global and local image sharpness estimation. IEEE Signal Processing Letters. 2012, 19(7): 423−426
    [15] 王义文, 刘献礼, 谢晖. 基于小波变换的显微图像清晰度评价函数及3-D自动调焦技术. 光学精密工程, 2006, 14(6): 1063−1069 doi: 10.3321/j.issn:1004-924X.2006.06.023

    Wang Yi-Wen, Liu Xian-Li, Xie Hui. A wavelet-based focus measure and 3-D autofocusing for microscope images. Optics and Precision Engineering, 2006, 14(6): 1063−1069 doi: 10.3321/j.issn:1004-924X.2006.06.023
    [16] Fan Z G, Chen S Q. Autofocus algorithm based on wavelet packet transform for infrared microscopy. In: Proceedings of the 3rd International Congress on Image and Signal Processing, Yantai, China: IEEE, 2010. 2510−2514
    [17] Zhang Z, Liu Y, Tan H. No-reference image sharpness assessment using scale and directional models. In: Proceedings of the 2018 IEEE International Conference on Multimedia & Expo. San Diego, CA, USA: IEEE, 2018. 1−6
    [18] Weijer J V D, Gevers T, Bagdanov A D. Boosting color saliency in image feature detection. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2005, 28(1): 150−156
    [19] Chen Y, Li D, Zhang J Q. Complementary color wavelet: A novel tool for the color image/video analysis and processing. IEEE Transactions on Circuits and Systems for Video Technology, 2017, 29(1): 12−27
    [20] MacAdam D L. Photometric relationships between complementary colors. Journal of the Optical Society of America, 1938, 28(4): 103−111 doi: 10.1364/JOSA.28.000103
    [21] Pridmore R W. Complementary colors theory of color vision: Physiology, color mixture, color constancy and color perception. Color Research & Application, 2011, 36(6): 394−412
    [22] Selesnick I W, Baraniuk R G, Kingsbury N C. The dual-tree complex wavelet transform. IEEE Signal Processing Magazine, 2005, 22(6): 123−151 doi: 10.1109/MSP.2005.1550194
    [23] Born M, Wolf E, Principles of optics. London: Pergamon, 1965.
    [24] Pentland P A. A new sense for depth of field. pattern analysis and machine intelligence. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1987, 9(4): 523−531
    [25] 陈扬, 李旦, 张建秋. 互补色小波域图像质量盲评价方法. 电子学报, 2019, 47(4): 775−783 doi: 10.3969/j.issn.0372-2112.2019.04.002

    Chen Yang, Li Dan, Zhang Jian-Qiu. Blind image quality assessment with complementary color wavelet transform. Acta Electronica Sinica, 2019, 47(4): 775−783 doi: 10.3969/j.issn.0372-2112.2019.04.002
    [26] 陈扬, 张建秋. 互补色小波域自然场景统计显著图模型. 微电子学与计算机, 2019, 36(3): 17−22, 27

    Chen Yang, Zhang Jian-Qiu. A naturel scene statistical saliency map model in complementary color wavelet domain. Microelectronics & Computer, 2019, 36(3): 17−22, 27
    [27] Ninassi A, Autrusseau F, Callet P L. Pseudo no reference image quality metric using perceptual data hiding. In: Proceedings of the International Society for Optical and Engineering. California, USA: SPIE, 2006.
    [28] Timofte R, Agustsson E, Gool L V. NTIRE 2017 challenge on single image super-resolution: methods and results. In: Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW). Honolulu, HI, USA: IEEE, 2017. 1110−1121
    [29] 苑津莎, 张冬雪, 李中. 基于改进阈值法的小波去噪算法研究. 华北电力大学学报, 2010, 37(5): 92−97

    Yuan Jin-Sha, Zhang Dong-Xue, Li Zhong. Wavelet denosing algorithm based on improved thresholding method. Joumal of North China Electric Power University, 2010, 37(5): 92−97
    [30] Sendur L, Selesnick I W. Bivariate shrinkage functions for wavelet-based denoising exploiting interscale dependency. IEEE Transactions on Signal Processing, 2002, 50(11): 2744−2756 doi: 10.1109/TSP.2002.804091
    [31] Sheikh H R, Wang Z, Cormack L, et al. LIVE image quality assessment database release 2(2005)[Online], available: http://live.ece.utexas.edu/research/quality/, September1, 2011
    [32] 方开泰, 许建伦. 统计分布. 北京: 科学出版社, 1987. 277−282.

    Fang Kai-Tai, Xu Jian-Lun. Statistical Distribution. Beijing Science Press, 1987. 277−282
    [33] 刘兴宝, 袁道成. 基于纹理分析的小波变换图像清晰度评价方法研究. 仪器仪表学报, 2007, 28(8): 1508−1513

    Liu Xin-Bao, Yuan Dao-Cheng. Research on image definition criterion using wavelet transform based on the tex ture analysis. Chinese Journal of Scientific Instrument, 2007, 28(8): 1508−1513
  • 加载中
图(11) / 表(9)
计量
  • 文章访问数:  894
  • HTML全文浏览量:  126
  • PDF下载量:  129
  • 被引次数: 0
出版历程
  • 收稿日期:  2019-07-04
  • 录用日期:  2020-04-16
  • 网络出版日期:  2020-08-26
  • 刊出日期:  2020-08-26

目录

    /

    返回文章
    返回