Journal of Computer Applications ›› 2022, Vol. 42 ›› Issue (1): 239-244.DOI: 10.11772/j.issn.1001-9081.2021010134
• Multimedia computing and computer simulation • Previous Articles
Suyu WANG1,2, Jing YANG1,2(), Yue LI1,2
Received:
2021-01-26
Revised:
2021-04-28
Accepted:
2021-04-29
Online:
2022-01-11
Published:
2022-01-10
Contact:
Jing YANG
About author:
WANG Suyu, born in 1976, Ph. D., associate professor. Her research interests include image and video signal processing, computer vision.通讯作者:
杨静
作者简介:
王素玉(1976—),女,河北唐山人,副教授,博士,主要研究方向:图像与视频信号处理、计算机视觉CLC Number:
Suyu WANG, Jing YANG, Yue LI. Image super-resolution restoration algorithm based on information distillation network with dual attention mechanism[J]. Journal of Computer Applications, 2022, 42(1): 239-244.
王素玉, 杨静, 李越. 基于双注意力机制信息蒸馏网络的图像超分辨率复原算法[J]. 《计算机应用》唯一官方网站, 2022, 42(1): 239-244.
Add to citation manager EndNote|Ris|BibTeX
URL: http://www.joca.cn/EN/10.11772/j.issn.1001-9081.2021010134
算法 | Scale | Set5 | Set14 | BSD100 | Urban100 | ||||
---|---|---|---|---|---|---|---|---|---|
PSNR/dB | SSIM | PSNR/dB | SSIM | PSNR/dB | SSIM | PSNR/dB | SSIM | ||
/ | ×4 | 31.57 | 0.886 5 | 28.18 | 0.772 3 | 27.20 | 0.728 0 | 25.33 | 0.762 5 |
SE | ×4 | 31.77 | 0.889 3 | 28.23 | 0.772 8 | 27.40 | 0.728 9 | 25.38 | 0.763 0 |
CBAM | ×4 | 31.79 | 0.891 0 | 28.24 | 0.772 9 | 27.42 | 0.730 0 | 25.42 | 0.763 3 |
RAM | ×4 | 31.80 | 0.890 2 | 28.26 | 0.772 9 | 27.43 | 0.730 2 | 25.41 | 0.770 0 |
Tab. 1 Comparison of results of different attention modules
算法 | Scale | Set5 | Set14 | BSD100 | Urban100 | ||||
---|---|---|---|---|---|---|---|---|---|
PSNR/dB | SSIM | PSNR/dB | SSIM | PSNR/dB | SSIM | PSNR/dB | SSIM | ||
/ | ×4 | 31.57 | 0.886 5 | 28.18 | 0.772 3 | 27.20 | 0.728 0 | 25.33 | 0.762 5 |
SE | ×4 | 31.77 | 0.889 3 | 28.23 | 0.772 8 | 27.40 | 0.728 9 | 25.38 | 0.763 0 |
CBAM | ×4 | 31.79 | 0.891 0 | 28.24 | 0.772 9 | 27.42 | 0.730 0 | 25.42 | 0.763 3 |
RAM | ×4 | 31.80 | 0.890 2 | 28.26 | 0.772 9 | 27.43 | 0.730 2 | 25.41 | 0.770 0 |
算法 | Scale | Set5 | Set14 | BSD100 | Urban100 | ||||
---|---|---|---|---|---|---|---|---|---|
PSNR/dB | SSIM | PSNR/dB | SSIM | PSNR/dB | SSIM | PSNR/dB | SSIM | ||
RAM+L1 | ×4 | 31.80 | 0.890 2 | 28.26 | 0.772 9 | 27.43 | 0.730 2 | 25.41 | 0.770 0 |
RAM+(MS-SSIM) | ×4 | 31.77 | 0.890 3 | 28.26 | 0.773 3 | 27.40 | 0.730 1 | 25.40 | 0.770 3 |
RAM+((MS-SSIM)+L1) | ×4 | 31.79 | 0.890 3 | 28.28 | 0.773 2 | 27.42 | 0.730 2 | 25.41 | 0.768 7 |
Tab. 2 Comparison of results of different loss functions
算法 | Scale | Set5 | Set14 | BSD100 | Urban100 | ||||
---|---|---|---|---|---|---|---|---|---|
PSNR/dB | SSIM | PSNR/dB | SSIM | PSNR/dB | SSIM | PSNR/dB | SSIM | ||
RAM+L1 | ×4 | 31.80 | 0.890 2 | 28.26 | 0.772 9 | 27.43 | 0.730 2 | 25.41 | 0.770 0 |
RAM+(MS-SSIM) | ×4 | 31.77 | 0.890 3 | 28.26 | 0.773 3 | 27.40 | 0.730 1 | 25.40 | 0.770 3 |
RAM+((MS-SSIM)+L1) | ×4 | 31.79 | 0.890 3 | 28.28 | 0.773 2 | 27.42 | 0.730 2 | 25.41 | 0.768 7 |
算法 | Scale | Set5 | Set14 | BSD100 | Urban100 | ||||
---|---|---|---|---|---|---|---|---|---|
PSNR/dB | SSIM | PSNR/dB | SSIM | PSNR/dB | SSIM | PSNR/dB | SSIM | ||
RAM | ×4 | 31.80 | 0.890 2 | 28.26 | 0.772 9 | 27.43 | 0.730 2 | 25.41 | 0.770 0 |
(MS-SSIM)+L1 | ×4 | 31.59 | 0.891 0 | 28.25 | 0.772 6 | 27.26 | 0.730 2 | 25.35 | 0.762 9 |
RAM+((MS-SSIM)+L1) | ×4 | 31.79 | 0.890 3 | 28.28 | 0.773 2 | 27.42 | 0.730 2 | 25.41 | 0.768 7 |
Tab. 3 Ablation comparison of RAM/(MS-SSIM)+L1
算法 | Scale | Set5 | Set14 | BSD100 | Urban100 | ||||
---|---|---|---|---|---|---|---|---|---|
PSNR/dB | SSIM | PSNR/dB | SSIM | PSNR/dB | SSIM | PSNR/dB | SSIM | ||
RAM | ×4 | 31.80 | 0.890 2 | 28.26 | 0.772 9 | 27.43 | 0.730 2 | 25.41 | 0.770 0 |
(MS-SSIM)+L1 | ×4 | 31.59 | 0.891 0 | 28.25 | 0.772 6 | 27.26 | 0.730 2 | 25.35 | 0.762 9 |
RAM+((MS-SSIM)+L1) | ×4 | 31.79 | 0.890 3 | 28.28 | 0.773 2 | 27.42 | 0.730 2 | 25.41 | 0.768 7 |
算法 | Scale | Set5 | Set14 | BSD100 | Urban100 | ||||
---|---|---|---|---|---|---|---|---|---|
PSNR/dB | SSIM | PSNR/dB | SSIM | PSNR/dB | SSIM | PSNR/dB | SSIM | ||
Bicubic | ×2 | 33.66 | 0.929 9 | 30.24 | 0.868 8 | 29.56 | 0.843 1 | 26.88 | 0.840 3 |
×3 | 30.39 | 0.868 2 | 27.55 | 0.774 2 | 27.21 | 0.738 5 | 24.46 | 0.734 9 | |
×4 | 28.42 | 0.810 4 | 26.00 | 0.702 7 | 25.96 | 0.667 5 | 23.14 | 0.657 7 | |
SRCNN | ×2 | 36.66 | 0.954 2 | 32.42 | 0.906 3 | 31.34 | 0.887 9 | 29.50 | 0.894 6 |
×3 | 32.75 | 0.909 0 | 29.28 | 0.820 9 | 28.40 | 0.786 3 | 26.24 | 0.798 9 | |
×4 | 30.48 | 0.862 8 | 27.49 | 0.750 3 | 26.90 | 0.710 1 | 24.52 | 0.722 1 | |
VDSR | ×2 | 37.53 | 0.958 7 | 33.03 | 0.912 4 | 31.90 | 0.896 0 | 30.76 | 0.914 0 |
×3 | 33.66 | 0.921 3 | 29.77 | 0.831 4 | 28.82 | 0.797 6 | 27.14 | 0.827 9 | |
×4 | 31.35 | 0.883 8 | 28.01 | 0.767 4 | 27.29 | 0.725 1 | 25.18 | 0.752 4 | |
DRCN | ×2 | 37.63 | 0.958 4 | 33.06 | 0.910 8 | 31.85 | 0.894 7 | 30.76 | 0.914 7 |
×3 | 33.85 | 0.921 5 | 29.89 | 0.831 7 | 28.81 | 0.795 4 | 27.16 | 0.831 1 | |
×4 | 31.56 | 0.881 0 | 28.15 | 0.762 7 | 27.24 | 0.715 0 | 25.15 | 0.753 0 | |
LapSRN | ×2 | 37.52 | 0.958 1 | 33.08 | 0.910 9 | 31.80 | 0.894 9 | 30.41 | 0.911 2 |
×3 | 33.82 | 0.920 7 | 29.89 | 0.830 4 | 28.82 | 0.795 0 | 27.07 | 0.829 8 | |
×4 | 31.54 | 0.881 1 | 28.19 | 0.763 5 | 27.32 | 0.716 2 | 25.21 | 0.756 4 | |
IDN | ×2 | 37.83 | 0.960 0 | 33.30 | 0.914 8 | 32.08 | 0.898 5 | 31.27 | 0.919 6 |
×3 | 34.11 | 0.925 3 | 29.99 | 0.835 4 | 28.95 | 0.801 3 | 27.42 | 0.835 9 | |
×4 | 31.82 | 0.890 3 | 28.25 | 0.773 0 | 27.41 | 0.729 7 | 25.41 | 0.763 2 | |
本文算法 | ×2 | 37.85 | 0.958 2 | 33.35 | 0.915 4 | 32.07 | 0.899 3 | 31.29 | 0.919 4 |
×3 | 34.13 | 0.928 9 | 29.97 | 0.835 1 | 28.89 | 0.821 1 | 27.41 | 0.841 7 | |
×4 | 31.79 | 0.890 3 | 28.28 | 0.773 2 | 27.42 | 0.730 2 | 25.41 | 0.768 7 |
Tab. 4 Comparison results of six algorithms
算法 | Scale | Set5 | Set14 | BSD100 | Urban100 | ||||
---|---|---|---|---|---|---|---|---|---|
PSNR/dB | SSIM | PSNR/dB | SSIM | PSNR/dB | SSIM | PSNR/dB | SSIM | ||
Bicubic | ×2 | 33.66 | 0.929 9 | 30.24 | 0.868 8 | 29.56 | 0.843 1 | 26.88 | 0.840 3 |
×3 | 30.39 | 0.868 2 | 27.55 | 0.774 2 | 27.21 | 0.738 5 | 24.46 | 0.734 9 | |
×4 | 28.42 | 0.810 4 | 26.00 | 0.702 7 | 25.96 | 0.667 5 | 23.14 | 0.657 7 | |
SRCNN | ×2 | 36.66 | 0.954 2 | 32.42 | 0.906 3 | 31.34 | 0.887 9 | 29.50 | 0.894 6 |
×3 | 32.75 | 0.909 0 | 29.28 | 0.820 9 | 28.40 | 0.786 3 | 26.24 | 0.798 9 | |
×4 | 30.48 | 0.862 8 | 27.49 | 0.750 3 | 26.90 | 0.710 1 | 24.52 | 0.722 1 | |
VDSR | ×2 | 37.53 | 0.958 7 | 33.03 | 0.912 4 | 31.90 | 0.896 0 | 30.76 | 0.914 0 |
×3 | 33.66 | 0.921 3 | 29.77 | 0.831 4 | 28.82 | 0.797 6 | 27.14 | 0.827 9 | |
×4 | 31.35 | 0.883 8 | 28.01 | 0.767 4 | 27.29 | 0.725 1 | 25.18 | 0.752 4 | |
DRCN | ×2 | 37.63 | 0.958 4 | 33.06 | 0.910 8 | 31.85 | 0.894 7 | 30.76 | 0.914 7 |
×3 | 33.85 | 0.921 5 | 29.89 | 0.831 7 | 28.81 | 0.795 4 | 27.16 | 0.831 1 | |
×4 | 31.56 | 0.881 0 | 28.15 | 0.762 7 | 27.24 | 0.715 0 | 25.15 | 0.753 0 | |
LapSRN | ×2 | 37.52 | 0.958 1 | 33.08 | 0.910 9 | 31.80 | 0.894 9 | 30.41 | 0.911 2 |
×3 | 33.82 | 0.920 7 | 29.89 | 0.830 4 | 28.82 | 0.795 0 | 27.07 | 0.829 8 | |
×4 | 31.54 | 0.881 1 | 28.19 | 0.763 5 | 27.32 | 0.716 2 | 25.21 | 0.756 4 | |
IDN | ×2 | 37.83 | 0.960 0 | 33.30 | 0.914 8 | 32.08 | 0.898 5 | 31.27 | 0.919 6 |
×3 | 34.11 | 0.925 3 | 29.99 | 0.835 4 | 28.95 | 0.801 3 | 27.42 | 0.835 9 | |
×4 | 31.82 | 0.890 3 | 28.25 | 0.773 0 | 27.41 | 0.729 7 | 25.41 | 0.763 2 | |
本文算法 | ×2 | 37.85 | 0.958 2 | 33.35 | 0.915 4 | 32.07 | 0.899 3 | 31.29 | 0.919 4 |
×3 | 34.13 | 0.928 9 | 29.97 | 0.835 1 | 28.89 | 0.821 1 | 27.41 | 0.841 7 | |
×4 | 31.79 | 0.890 3 | 28.28 | 0.773 2 | 27.42 | 0.730 2 | 25.41 | 0.768 7 |
1 | DONG C, LOY C C, HE K M, et al. Image super-resolution using deep convolutional networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016, 38(2):295-307. 10.1109/tpami.2015.2439281 |
2 | HUI Z, WANG X M, GAO X B. Fast and accurate single image super-resolution via information distillation network[C]// Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2018:723-731. 10.1109/cvpr.2018.00082 |
3 | ZHAO H, GALLO O, FROSIO I, et al. Loss functions for image restoration with neural networks[J]. IEEE Transactions on Computational Imaging, 2017, 3(1):47-57. 10.1109/tci.2016.2644865 |
4 | ZHAO H, GALLO O, FROSIO I, et al. Loss functions for neural networks for image processing[EB/OL]. (2018-04-20) [2020-10-15].. 10.1109/tci.2016.2644865 |
5 | YANG J C, WRIGHT J, HUANG T S, et al. Image super-resolution via sparse representation[J]. IEEE Transactions on Image Processing, 2010, 19(11):2861-2873. 10.1109/tip.2010.2050625 |
6 | ZEYDE R, ELAD M, PROTTER M. On single image scale-up using sparse-representations[C]// Proceedings of the 2010 International Conference Curves and Surfaces, LNCS6920. Berlin: Springer, 2012: 711-730. |
7 | KIM J, LEE J K, LEE K M. Accurate image super-resolution using very deep convolutional networks[C]// Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2016: 1646-1654. 10.1109/cvpr.2016.182 |
8 | NOH H, HONG S, HAN B. Learning deconvolution network for semantic segmentation[C]// Proceedings of the 2015 IEEE International Conference on Computer Vision. Piscataway: IEEE, 2015: 1520-1528. 10.1109/iccv.2015.178 |
9 | LIM B, SON S, KIM H, et al. Enhanced deep residual networks for single image super-resolution[C]// Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops. Piscataway: IEEE, 2017: 1132-1140. 10.1109/cvprw.2017.151 |
10 | KIM J H, CHOI J H, CHEON M, et al. RAM: residual attention module for single image super-resolution[EB/OL]. (2018-11-29) [2020-10-15].. |
11 | MARTIN D, FOWLKES C, TAL D, et al. A database of human segmented natural images and its application to evaluating segmentation algorithms and measuring ecological statistics[C]// Proceedings of the 8th IEEE International Conference on Computer Vision. Piscataway: IEEE, 2001:416-423. 10.1109/iccv.2001.937655 |
12 | HU J, SHEN L, ALBANIE S, et al. Squeeze-and-excitation networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2020, 42(8):2011-2023. 10.1109/tpami.2019.2913372 |
13 | WOO S, PARK J, LEE J Y. CBAM: convolutional block attention module[C]// Proceedings of the 2018 European Conference on Computer Vision, LNCS11211. Cham: Springer, 2018:3-19. |
14 | DONG C, LOY C C, HE K M, et al. Learning a deep convolutional network for image super-resolution[C]// Proceedings of the 2014 European Conference on Computer Vision, LNCS8692. Cham: Springer, 2014:184-199. |
15 | KIM J, LEE J K, LEE K M. Deeply-recursive convolutional network for image super-resolution[C]// Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2016:1637-1645. 10.1109/cvpr.2016.181 |
16 | LAI W S, HUANG J B, AHUJA N, et al. Deep Laplacian pyramid networks for fast and accurate super-resolution[C]// Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2017:5835-5843. 10.1109/cvpr.2017.618 |
Viewed | ||||||
Full text |
|
|||||
Abstract |
|
|||||