Super-resolution image reconstruction based on convolutional sparse coding and generative adversarial networks

Fig. 8. $2 \times$ super-resolution image reconstruction results of each algorithm.（a）High-resolution original images；（b）Bicubic；（c）SRGAN；（d）EDSR；（e）ESRGAN；（f）Ours.

Download full size

View in Article

Fig. 9. $4 \times$ super-resolution image reconstruction results of each algorithm.（a）High-resolution original images；（b）Bicubic；（c）SRGAN；（d）EDSR；（e）ESRGAN；（f）Ours.

Download full size

View in Article

Table 1. Average PSNR，SSIM and LPIPS values for each algorithm for $\times 2$ magnification

View table

View in Article

Table 1. Average PSNR，SSIM and LPIPS values for each algorithm for $\times 2$ magnification

算法	指标	Set5	Set14	BSD100	Urban100
Bicubic	PSNR	30.420 3	26.802 5	26.472 3	24.158 3
	SSIM	0.899 2	0.852 4	0.796 3	0.787 7
	LPIPS	0.172 1	0.142 6	0.230 7	0.194 3
SRGAN	PSNR	30.903 0	28.137 0	27.845 5	25.819 6
	SSIM	0.900 6	0.791 7	0.832 6	0.812 1
	LPIPS	0.027 6	0.054 2	0.094 5	0.071 8
EDSR	PSNR	31.900 2	28.258 3	28.101 6	26.001 2
	SSIM	0.896 8	0.828 7	0.828 7	0.810 1
	LPIPS	0.037 1	0.061 2	0.100 4	0.079 8
ESRGAN	PSNR	32.230 1	28.923 9	28.317 0	26.702 5
	SSIM	0.901 1	0.845 5	0.835 0	0.823 1
	LPIPS	0.020 7	0.058 6	0.083 6	0.068 8
Ours	PSNR	32.074 3	29.519 1	28.632 3	27.635 0
	SSIM	0.907 1	0.864 0	0.857 9	0.875 9
	LPIPS	0.016 3	0.041 4	0.068 3	0.062 1

Table 2. Average PSNR，SSIM and LPIPS values for each algorithm for $\times 4$ magnification

View table

View in Article

Table 2. Average PSNR，SSIM and LPIPS values for each algorithm for $\times 4$ magnification

算法	指标	Set5	Set14	BSD100	Urban100
Bicubic	PSNR	26.998 1	25.251 3	23.125 1	22.951 7
	SSIM	0.765 1	0.660 8	0.703 1	0.691 6
	LPIPS	0.310 2	0.392 7	0.421 9	0.312 1
SRGAN	PSNR	27.316 5	26.572 6	25.006 9	22.458 3
	SSIM	0.793 8	0.702 8	0.727 8	0.693 3
	LPIPS	0.169 8	0.235 6	0.241 4	0.221 4
EDSR	PSNR	27.702 6	26.231 9	26.031 5	23.052 6
	SSIM	0.800 7	0.761 2	0.753 7	0.701 5
	LPIPS	0.172 4	0.211 9	0.264 6	0.251 7
ESRGAN	PSNR	29.502 1	27.045 2	26.219 2	23.173 9
	SSIM	0.841 6	0.771 0	0.732 1	0.721 2
	LPIPS	0.153 2	0.181 2	0.191 8	0.190 2
Ours	PSNR	30.345 2	27.991 8	27.302 4	24.235 9
	SSIM	0.851 0	0.817 3	0.764 1	0.772 5
	LPIPS	0.113 0	0.163 0	0.185 4	0.170 9

Tools

Get Citation

Copy Citation Text

Jun-sen DU, Jie-long GUO, Hui YU, Xian WEI. Super-resolution image reconstruction based on convolutional sparse coding and generative adversarial networks[J]. Chinese Journal of Liquid Crystals and Displays, 2023, 38(10): 1423

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites

Paper Information

Category: Research Articles

Received: Dec. 6, 2022

Accepted: --

Published Online: Oct. 25, 2023

The Author Email: Jie-long GUO (gjl@fjirsm.ac.cn)

DOI:10.37188/CJLCD.2022-0406

Topics

Table 1. Average PSNR，SSIM and LPIPS values for each algorithm for ×2 magnification

Table 1. Average PSNR，SSIM and LPIPS values for each algorithm for ×2 magnification

Table 2. Average PSNR，SSIM and LPIPS values for each algorithm for ×4 magnification

Table 2. Average PSNR，SSIM and LPIPS values for each algorithm for ×4 magnification

Table 1. Average PSNR，SSIM and LPIPS values for each algorithm for $\times 2$ magnification

Table 1. Average PSNR，SSIM and LPIPS values for each algorithm for $\times 2$ magnification

Table 2. Average PSNR，SSIM and LPIPS values for each algorithm for $\times 4$ magnification

Table 2. Average PSNR，SSIM and LPIPS values for each algorithm for $\times 4$ magnification