Chinese Journal of Liquid Crystals and Displays, Volume. 39, Issue 10, 1391(2024)
Lightweight image super-resolution combining residual learning and layer attention
Figures & Tables(12)
Fig. 5. Visual results of RLAN（Ours）compared with other advanced SR models（×4）
Table 1. Metric results of RLAN（Ours）compared with other advanced SR models（×2）
View tableView in ArticleTable 1. Metric results of RLAN（Ours）compared with other advanced SR models（×2）
Model Params/k Metric Set5 Set14 BSD100 Urban100 VDSR 666 PSNR/SSIM 37.53/0.958 7 33.05/0.912 7 31.90/0，896 0 30.77/0.914 1 DRCN 1 774 PSNR/SSIM 37.63/0.958 8 33.04/0.911 8 31.85/0.894 2 30.75/0.913 3 SMSR 985 PSNR/SSIM 38.00/0.960 1 33.64/0.917 9 32.17/0.899 0 32.19/0.928 4 LatticeNet-CL 756 PSNR/SSIM 38.06/ 0.960 8 33.70/ 0.918 8 32.21/0.900 0 32.29/0.929 1 RLFN 543 PSNR/SSIM 38.07 /0.960 733.72 /0.918 732.22 /0.900 2 32.33 /0.930 2 RLAN（Ours） 373 PSNR/SSIM 38.29 /0.961 8 34.26 /0.920 1 32.81 /0.901 5 32.61 /0.929 4
Table 2. Metric results of RLAN（Ours）compared with other advanced SR models（×3）
View tableView in ArticleTable 2. Metric results of RLAN（Ours）compared with other advanced SR models（×3）
Model Params/k Metric Set5 Set14 BSD100 Urban100 VDSR 666 PSNR/SSIM 33.66/0.921 3 29.77/0.831 4 28.80/0.797 4 27.14/0.827 9 DRCN 1 774 PSNR/SSIM 33.82/0.922 6 29.76/0.831 1 28.82/0.796 7 27.15/0.827 7 SMSR 993 PSNR/SSIM 34.40/0.927 0 30.33/0.841 2 29.10/0.805 0 28.25/0.853 6 LatticeNet-CL 765 PSNR/SSIM 34.46/0.927 5 30.37/0.842 2 29.12/0.805 4 28.23/0.852 5 RLFN 543 PSNR/SSIM 34.52 /0.928 1 30.41 /0.843 8 29.15 /0.805 7 28.39/0.856 9 RLAN（Ours） 373 PSNR/SSIM 34.71 /0.929 3 30.57 /0.846 1 29.23 /0.805 9 28.68 /0.860 1
Table 3. Metric results of RLAN（Ours）compared with other advanced SR models（×4）
View tableView in ArticleTable 3. Metric results of RLAN（Ours）compared with other advanced SR models（×4）
Model Params/k Metric Set5 Set14 BSD100 Urban100 VDSR 666 PSNR/SSIM 31.35/0.883 8 28.01/0.767 4 27.29/0.725 1 25.18/0.752 4 DRCN 1 774 PSNR/SSIM 31.53/0.885 4 28.02/0.767 0 27.23/0.723 3 25.14/0.751 0 SMSR 1 006 PSNR/SSIM 32.12/0.893 2 28.55/0.780 6 27.55/0.735 1 26.11/0.785 2 LatticeNet-CL 777 PSNR/SSIM 32.27 /0.894 428.61/ 0.781 9 27.57/ 0.736 5 26.16/0.785 5 RLFN 543 PSNR/SSIM 32.24/ 0.894 9 28.63 /0.781 027.58 /0.736 4 26.17 /0.786 7 RLAN（Ours） 373 PSNR/SSIM 32.58 /0.899 4 28.76 /0.782 1 28.02 /0.735 827.06 /0.788 4
Table 4. Experimental results of structural validity of RLFB（×4）
View tableView in ArticleTable 4. Experimental results of structural validity of RLFB（×4）
Model Params/k PSNR RLAN_ca 394 28.65 RLAN_pconv 178 28.11 RLAN 373 28.76
Table 5. Experimental results of the effectiveness of pixel attention reconstruction block（×4）
View tableView in ArticleTable 5. Experimental results of the effectiveness of pixel attention reconstruction block（×4）
Model Params/k PSNR RLAN_ps 378 28.58 RLAN 373 28.76
Table 6. Experimental results of pixel attention reconstruction block and convolution（×4）
View tableView in ArticleTable 6. Experimental results of pixel attention reconstruction block and convolution（×4）
Model Params/k PSNR RLFN_4 378 28.60 RLFN_6 543 28.63 RLFN_4p 373 28.65
Tools
Get Citation
Copy Citation Text
Difan WU, Xuande ZHANG. Lightweight image super-resolution combining residual learning and layer attention[J]. Chinese Journal of Liquid Crystals and Displays, 2024, 39(10): 1391
Paper Information
Category:
Received: Feb. 18, 2024
Accepted: --
Published Online: Nov. 13, 2024
The Author Email: Xuande ZHANG (zhangxuande@sust.edu.cn)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20