Acta Physica Sinica, Volume. 69, Issue 14, 148702-1(2020)

Objective assessment of image quality based on image content contrast perception

Jun-Cai Yao1...23,* and Jing Shen1 |Show fewer author(s)
Author Affiliations
  • 1School of Computer Engineering, Nanjing Institute of Technology, Nanjing 211167, China
  • 2School of Physics and Telecommunication Engineering, Shaanxi University of Technology, Hanzhong 723000, China
  • 3School of Information and Communications Engineering, Xi’an Jiaotong University, Xi’an 710049, China
  • show less
    Figures & Tables(13)
    The architecture of the proposed IQA method.
    Scatter plots between the subjective and objective IQA results of images in four databases: (a) LIVE; (b) CSIQ; (c) TID2008; (d) TID2013
    IQA results of the gray and monochrome images in IVC database by the proposed model.
    Comparing the accuracy of the proposed model with those of the existing 7 models based on the IQA results in TID2008 database: (a) PSNR-TID2008; (b) VSNR-TID2008; (c) SSIM-TID2008; (d) FSIMc-TID2008; (e) VSI-TID2008; (f) GMSD-TID2008; (g) MAD-TID2008; (h) MPCC-TID2008.
    Comparison of the complexity of 8 IQA models based on the IQA running time per 10 images.
    Accuracy comparisons among 8 IQA metrics based on PLCC of IQA results from 28 types of distortion images in three databases: (a) CSIQ; (b) LIVE; (c) TID2008.
    Scatter plots of the IQA results of 6 kinds of distorted images in CSIQ database evaluating by the proposed IQA model: (a) awgn; (b) jpeg; (c) jpeg2k; (d) fnoise; (e) blur; (f) contrast.
    Scatter plots of the IQA results of 5 kinds of distorted images in LIVE database evaluating by the proposed IQA model: (a) jpeg2k; (b) jpeg; (c) WN; (d) gblur; (e) fastfading.
    Scatter plots of the IQA results of 17 kinds of distorted images in TID2008 database evaluating by the proposed IQA model: (a) AGN; (b) ANCC; (c) SCN; (d) MN; (e) HFN; (f) IN; (g) QN; (h) GB; (i) ID; (j) JPEG; (k) JPEG2k; (l) JPEGtrans; (m) JPEG2ktrans; (n) NEPN; (o) LBWD; (p) MS; (q) CC.
    Scatter plots of the IQA results of 24 kinds of distorted images in TID2013 database evaluating by the proposed IQA model: (a) AGN; (b) NCC; (c) SCN; (d) MN; (e) HFN; (f) IN; (g) QN; (h) GB; (i) ID; (j) JPEG; (k) JPEG2k; (l) JPEGtrans; (m) JPEG2ktrans; (n) NEPN; (o) LBWD; (p) MS; (q) CC; (r) CCS; (s) MGN; (t) CN; (u) LCN; (v) CQWD; (w) CA; (x) SSR.
    • Table 1. Calculated 4 correlation parameters between the subjective and objective IQA scores of images in 4 databases.

      View table
      View in Article

      Table 1. Calculated 4 correlation parameters between the subjective and objective IQA scores of images in 4 databases.

      数据库LIVE(779)CSIQ(866)TID2008(1700)TID2013(3000)加权
      PLCC0.96220.95860.87780.86160.8915
      SROCC0.96600.95690.88310.84520.8854
      RMSE7.43970.07470.64270.6293
      OR0.15310.26900.12870.1198
    • Table 2.

      Comparing the accuracy of the proposed model with those of the existing 7 models based on the IQA results in CSIQ, LIVE, and TID2013 databases.

      基于CSIQ, LIVE和TID2013数据库中的图像IQA结果比较所提模型与现有7个模型的精度

      View table
      View in Article

      Table 2.

      Comparing the accuracy of the proposed model with those of the existing 7 models based on the IQA results in CSIQ, LIVE, and TID2013 databases.

      基于CSIQ, LIVE和TID2013数据库中的图像IQA结果比较所提模型与现有7个模型的精度

      数据库参数PSNRVSNRSSIMFSIMcVSIGMSDMADMPCC
      CSIQPLCC0.80000.80020.86130.91920.92790.95410.95020.9587
      SROCC0.80580.81060.87560.93100.94230.95700.94660.9569
      RMSE0.15750.15750.13340.10340.09790.07860.08180.0748
      OR0.42200.38320.35350.30410.28730.27420.28290.2738
      LIVEPLCC0.87230.92310.94490.96130.94820.96030.96750.9620
      SROCC0.87560.92740.94790.96450.95240.96030.96690.9660
      RMSE13.359710.50598.94557.52968.68167.62146.90737.4598
      OR0.21790.21510.18650.16270.18530.16430.15290.1606
      TID2013PLCC0.70620.74020.78950.87690.90000.85530.82670.8648
      SROCC0.69170.73160.74170.85100.89650.80440.78070.8452
      RMSE0.88870.83920.76080.59590.54040.64230.69750.6224
      OR0.16360.15520.14270.11320.10450.12420.13230.1179
    • Table 3. [in Chinese]

      View table
      View in Article

      Table 3. [in Chinese]

      失真类别PSNRVSNRSSIMFSIMcVSIGMSDMADMPCC
      1 Additive Gaussian noise(AGN)0.95520.83190.86850.91520.95270.95030.88970.8706
      2 Noise in color comp. (NCC)0.92560.78140.80500.88730.91720.91180.84380.8324
      3 Spatially correl. noise (SCN)0.95250.81050.86210.89890.94720.93910.90080.7457
      4 Masked noise (MN)0.87070.77150.82190.84920.82030.75470.80090.6943
      5 High frequency noise (HFN)0.97310.90610.90810.94750.96550.95670.92330.9090
      6 Impulse noise (IN)0.88870.74420.74150.81710.86350.75720.32060.7408
      7 Quantization noise (QN)0.88800.83840.87020.87940.87470.91100.85710.8122
      8 Gaussian blur (GB)0.91690.94370.96340.95440.95510.90990.93570.9252
      9 Image denoising (ID)0.96400.94630.95890.96520.97070.97590.96450.9594
      10 JPEG compression (JPEG)0.91670.93860.95510.97540.98580.98430.96380.9509
      11 JPEG2000 compression (JPEG2 K)0.91700.95130.96580.97540.98450.98120.97400.9452
      12 JPEG transm. errors (JPEG trans.)0.81040.85970.91810.91760.94570.90790.90010.8805
      13 JPEG2000 transm. errors (JPEG2K trans)0.90020.84350.88010.89290.91920.90850.88380.8699
      14 Non ecc. patt. noise (NEPN)0.67460.67740.77730.80680.81620.81330.86080.8132
      15 Local block-wise dist. (LBWD)0.24100.36320.60220.55420.49840.65200.41870.6845
      16 Mean shift (MS)0.80560.51600.80190.78690.80210.77070.69340.7720
      17 Contrast change (CC)0.58110.42510.60260.72660.69740.71110.31990.8108
      18 Change of color saturation (CSS)0.32940.41840.45900.82280.80520.42340.28460.7583
      19 Multipl. Gauss. noise (MGN)0.92040.77300.78960.86600.91360.89110.85290.8759
      20 Comfort noise (CN)0.87020.90160.90220.94630.95460.95620.94440.8476
      21 Lossy compr. of noisy (LCN)0.94290.89600.91740.95640.96360.97030.95620.7889
      22 Image color quant. w. dither (CQWD)0.93080.87730.86190.89110.89630.91920.87790.8721
      23 Chromatic aberrations (CA)0.95560.95920.97700.97940.97480.97370.96960.9473
      24 Sparse sampl. and reconstr. (SSR)0.92960.94770.96670.97760.98080.98490.97660.9349
      Max0.97310.95920.97700.97940.98580.98490.97660.9594
      Min0.24100.36320.45900.55420.49840.42340.28460.6845
      波动范围宽度0.73210.59590.51810.42520.48730.56140.69200.2750
      所有整体精度0.70620.74020.78950.87690.90000.85530.82670.8648
    Tools

    Get Citation

    Copy Citation Text

    Jun-Cai Yao, Jing Shen. Objective assessment of image quality based on image content contrast perception[J]. Acta Physica Sinica, 2020, 69(14): 148702-1

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category:

    Received: Mar. 4, 2020

    Accepted: --

    Published Online: Dec. 28, 2020

    The Author Email:

    DOI:10.7498/aps.69.20200335

    Topics