Optics and Precision Engineering, Volume. 31, Issue 10, 1509(2023)
Full reference image quality assessment based on color appearance-based phase consistency
Figures & Tables(11)
Fig. 1. Performance comparison chart of CPC algorithm and traditional PC algorithm
Fig. 5. SROCC of different
Fig. 6. SROCC of different
Fig. 7. Scatter diagram of subjective MOS and model prediction scores on TID 2013 database
Table 1. IQA benchmark database
View tableView in ArticleTable 1. IQA benchmark database
Database Source images Distorted images Distortion types Observers TID2013 25 3 000 24 971 TID2008 25 1 700 17 838 CSIQ 30 866 6 35 LIVE 29 779 5 161
Table 2. Performance comparison of different IQA models in four databases
View tableView in ArticleTable 2. Performance comparison of different IQA models in four databases
Database Criteria SSIM VIF MAD FSIMc GMSD VSI GSM RVISM CVSS CAGS Proposed TID2013 SROCC 0.741 7 0.676 9 0.780 7 0.851 0 0.804 4 0.896 5 0.794 6 0.675 7 0.806 9 0.831 6 0.859 2 KROCC 0.558 8 0.514 7 0.603 5 0.666 5 0.633 9 0.718 3 0.625 5 0.514 6 0.633 1 0.646 9 0.673 8 PLCC 0.789 5 0.772 0 0.826 7 0.876 9 0.858 3 0.900 0 0.846 4 0.782 5 0.840 6 0.844 5 0.878 1 RMSE 0.760 8 0.788 0 0.697 5 0.595 9 0.636 0 0.540 4 0.660 3 0.771 9 0.671 5 0.663 9 0.593 2 TID2008 SROCC 0.774 9 0.749 1 0.834 0 0.884 0 0.890 7 0.897 9 0.850 4 0.737 5 0.896 1 0.823 1 0.891 6 KROCC 0.576 8 0.586 0 0.644 5 0.699 1 0.709 2 0.712 3 0.659 6 0.562 8 0.958 0 0.628 9 0.704 0 PLCC 0.773 2 0.808 4 0.830 8 0.876 2 0.878 8 0.876 2 0.842 2 0.795 4 0.595 6 0.809 1 0.886 8 RMSE 0.851 1 0.789 9 0.746 8 0.646 8 0.640 4 0.646 6 0.723 5 0.813 3 0.721 5 0.788 6 0.620 1 CSIQ SROCC 0.875 6 0.919 5 0.946 6 0.931 0 0.957 0 0.942 3 0.910 8 0.897 9 0.958 0 0.919 8 0.950 5 KROCC 0.690 7 0.753 7 0.797 0 0.769 0 0.812 9 0.785 7 0.737 4 0.723 4 0.840 6 0.748 7 0.800 9 PLCC 0.861 3 0.927 7 0.950 2 0.919 2 0.954 1 0.927 9 0.896 4 0.923 6 0.958 9 0.901 4 0.953 7 RMSE 0.133 4 0.098 0 0.081 8 0.103 4 0.078 6 0.097 9 0.116 4 0.100 7 0.074 5 0.113 7 0.079 0 LIVE SROCC 0.947 9 0.963 6 0.966 9 0.964 5 0.960 3 0.952 4 0.956 1 0.960 0 0.967 2 0.973 4 0.965 3 KROCC 0.796 3 0.828 2 0.842 1 0.836 3 0.826 8 0.805 8 0.815 0 0.820 3 0.840 6 0.865 8 0.835 6 PLCC 0.944 9 0.960 4 0.967 5 0.961 3 0.959 5 0.948 2 0.951 2 0.957 0 0.965 1 0.964 0 0.961 6 RMSE 8.945 5 7.613 7 6.907 3 7.529 6 7.693 7 8.681 6 8.432 7 7.927 4 7.157 3 8.325 1 7.498 3 权重平均 SROCC 0.794 2 0.764 6 0.840 5 0.884 7 0.854 4 0.910 0 0.845 2 0.763 8 0.875 1 0.858 8 0.896 4 KROCC 0.610 8 0.604 9 0.670 2 0.710 1 0.702 2 0.736 6 0.673 2 0.600 6 0.711 5 0.682 8 0.730 0 PLCC 0.814 0 0.826 1 0.861 9 0.892 8 0.889 2 0.903 3 0.865 0 0.830 7 0.889 3 0.857 5 0.901 0 直接平均 SROCC 0.835 0 0.827 3 0.882 1 0.907 6 0.903 1 0.922 3 0.878 0 0.817 8 0.908 0 0.887 0 0.916 7 KROCC 0.655 7 0.670 7 0.721 8 0.742 7 0.745 7 0.755 5 0.709 4 0.655 3 0.753 1 0.722 6 0.753 6 PLCC 0.842 2 0.867 1 0.893 8 0.908 4 0.912 7 0.913 1 0.884 1 0.864 6 0.915 2 0.879 8 0.920 1
Table 3. SROCC values of IQA model under different distortion types
View tableView in ArticleTable 3. SROCC values of IQA model under different distortion types
Database Type SSIM VIF MAD FSIMc GMSD VSI GSM RVISM CVSS CAGS Proposed TID2013 AGN 0.867 1 0.899 4 0.884 3 0.910 1 0.946 2 0.946 0 0.906 4 0.864 5 0.940 1 0.935 9 0.932 5 ANC 0.772 6 0.829 9 0.801 9 0.853 7 0.868 4 0.870 5 0.817 5 0.812 4 0.863 9 0.865 3 0.843 4 SCN 0.851 5 0.883 5 0.891 1 0.890 0 0.935 0 0.936 7 0.915 8 0.841 2 0.907 7 0.927 6 0.911 9 MN 0.776 7 0.845 0 0.738 0 0.809 4 0.707 5 0.769 7 0.729 3 0.822 8 0.771 5 0.752 6 0.781 5 HFN 0.863 4 0.897 2 0.887 6 0.904 0 0.916 2 0.920 0 0.886 9 0.883 7 0.909 7 0.915 9 0.908 0 IN 0.750 3 0.853 7 0.276 9 0.825 1 0.763 7 0.874 1 0.796 5 0.882 3 0.745 7 0.836 1 0.781 4 QN 0.865 7 0.785 4 0.851 4 0.880 7 0.904 9 0.874 8 0.884 1 0.751 9 0.886 9 0.871 8 0.899 3 GB 0.966 8 0.965 0 0.931 9 0.955 1 0.911 3 0.961 2 0.968 9 0.974 2 0.934 8 0.961 4 0.926 8 DEN 0.925 4 0.891 1 0.925 2 0.933 0 0.952 5 0.948 4 0.943 2 0.894 7 0.942 7 0.946 6 0.946 6 JPEG 0.920 0 0.919 2 0.921 7 0.933 9 0.950 7 0.954 1 0.928 4 0.930 6 0.952 1 0.958 5 0.946 3 JP2K 0.946 8 0.951 6 0.951 1 0.958 9 0.965 7 0.970 6 0.960 2 0.952 1 0.958 7 0.962 0 0.954 7 JGTE 0.849 3 0.840 9 0.828 3 0.861 0 0.840 3 0.921 6 0.851 2 0.843 4 0.861 3 0.864 4 0.883 2 J2TE 0.882 8 0.876 1 0.878 8 0.891 9 0.913 6 0.922 8 0.918 2 0.885 4 0.885 1 0.925 0 0.880 5 NEPN 0.782 1 0.772 0 0.831 5 0.793 7 0.814 0 0.806 0 0.813 0 0.750 2 0.820 1 0.783 3 0.831 8 Block 0.572 0 0.530 6 0.281 2 0.553 2 0.662 5 0.171 3 0.641 8 0.603 7 0.515 2 0.601 5 0.662 0 MS 0.775 2 0.627 6 0.645 0 0.748 7 0.735 1 0.770 0 0.787 5 0.607 9 0.715 0 0.744 1 0.710 7 CTC 0.377 5 0.838 6 0.197 2 0.467 9 0.323 5 0.475 4 0.485 7 0.152 6 0.294 0 0.451 4 0.314 7 CCS 0.414 1 0.309 9 0.057 5 0.810 0 0.294 8 0.274 8 0.357 8 0.399 4 0.261 4 0.371 1 0.584 0 MGN 0.780 3 0.846 8 0.840 9 0.856 9 0.888 6 0.911 7 0.834 8 0.823 3 0.879 9 0.870 0 0.868 8 CN 0.856 6 0.894 6 0.906 4 0.924 3 0.929 8 0.912 1 0.912 4 0.898 4 0.935 1 0.916 8 0.934 5 LCNI 0.905 7 0.920 4 0.944 3 0.948 5 0.962 9 0.956 4 0.956 3 0.915 6 0.962 9 0.957 4 0.965 6 ICQD 0.854 2 0.841 4 0.874 5 0.881 5 0.910 2 0.883 9 0.897 3 0.803 6 0.910 8 0.906 0 0.906 1 CHA 0.877 5 0.884 8 0.831 0 0.892 5 0.853 0 0.890 6 0.882 3 0.915 4 0.852 3 0.876 8 0.857 7 SSR 0.946 1 0.935 3 0.956 7 0.957 6 0.968 3 0.962 8 0.966 8 0.943 9 0.960 5 0.958 0 0.960 3 CSIQ AGWN 0.897 4 0.957 5 0.954 1 0.935 9 0.968 0 0.963 6 0.944 0 0.937 7 0.967 0 0.965 2 0.949 0 JPEG 0.954 6 0.970 5 0.961 5 0.966 4 0.965 0 0.961 8 0.963 2 0.959 9 0.968 9 0.957 3 0.968 2 JP2K 0.960 6 0.967 2 0.975 2 0.970 4 0.972 0 0.969 4 0.964 8 0.969 3 0.977 7 0.954 5 0.973 0 AGPN 0.892 2 0.951 1 0.957 0 0.937 0 0.950 0 0.963 8 0.938 7 0.931 9 0.951 6 0.949 2 0.932 6 GB 0.960 9 0.974 5 0.960 2 0.972 9 0.971 0 0.967 9 0.958 9 0.971 5 0.978 9 0.957 4 0.975 4 GCD 0.792 2 0.934 5 0.920 7 0.943 8 0.904 0 0.950 4 0.935 4 0.879 2 0.932 4 0.927 3 0.927 4 LIVE JP2K 0.961 4 0.969 6 0.967 6 0.972 4 0.971 0 0.960 4 0.970 0 0.962 2 0.971 9 0.982 2 0.970 7 JPEG 0.976 4 0.984 6 0.976 4 0.984 0 0.978 0 0.976 1 0.977 8 0.982 2 0.983 6 0.983 6 0.980 4 AWGN 0.969 4 0.985 8 0.984 4 0.971 6 0.974 0 0.983 5 0.977 4 0.969 2 0.980 9 0.983 7 0.966 9 GB 0.951 7 0.972 8 0.946 5 0.970 8 0.957 0 0.952 7 0.951 8 0.964 3 0.966 2 0.964 1 0.968 9 FF 0.955 6 0.965 0 0.956 9 0.951 9 0.942 0 0.943 0 0.940 2 0.958 1 0.959 2 0.963 3 0.968 0
Table 4. Comparison of efficiency of different IQA models
View tableView in ArticleTable 4. Comparison of efficiency of different IQA models
IQA index Time cost /s IQA index Time cost/s SSIM 0.278 RVSIM 0.939 PSNR 0.156 MAD 1.256 CAGS 0.483 CPCCs 0.531 VSI 0.536
Tools
Get Citation
Copy Citation Text
Benchi JIANG, Shilei BIAN, Chenyang SHI, Lulu WU. Full reference image quality assessment based on color appearance-based phase consistency[J]. Optics and Precision Engineering, 2023, 31(10): 1509
Paper Information
Category: Information Sciences
Received: May. 14, 2022
Accepted: --
Published Online: Jul. 4, 2023
The Author Email:
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20