Cascade residual-optimized image super-resolution reconstruction in Transformer network

Table 1. Simulation environment and parameter setting
View table
View in Article
Table 1. Simulation environment and parameter setting
仿真环境参数值
中央处理器CPU Intel（R） Xeon（R） Silver 4210R
GPU Quadro RTX 5000
内存 16 GB
训练框架 Pytorch1.12.1
初始学习率 0.000 2
优化器 Adam
迭代次数Epoch 3 000

Table 2. Validity verification of different modules
View table
View in Article
Table 2. Validity verification of different modules
模型
编号
全局
级联
CA
感知
模块
局部
级联
PSNR/SSIM
Baseline × × × × 32.14/0.893 9
1 √ × × × 32.16/0.893 6
2 √ √ × × 32.27/0.894 7
3 √ √ √ × 32.37/0.896 2
ours √ √ √ √ 32.46/0.897 3

Table 3. Network performance with different number of modules
View table
View in Article
Table 3. Network performance with different number of modules
名称模块数参数量/10³ 浮点数计算量/10⁹ PSNR/SSIM
CR-Transformer 2 1，141 122 32.31/0.895 4
3 1，675 165 32.46/0.897 3
4 2，214 208 32.39/0.896 5
5 2，776 251 32.37/0.896 2

Table 4. Comparison of PSNR/SSIM values of different algorithms under 2× reconstruction

View table

View in Article

Table 4. Comparison of PSNR/SSIM values of different algorithms under 2× reconstruction

尺度	方法	Set5		Set14		BSD100		Urban100		Manga109
尺度	方法	PSNR	SSIM	PSNR	SSIM	PSNR	SSIM	PSNR	SSIM	PSNR	SSIM
×2	Bicubic	33.66	0.929 9	30.24	0.868 8	29.56	0.843 1	26.88	0.840 3	30.80	0.933 9
	SRCNN［4］	36.66	0.954 2	32.45	0.906 7	31.36	0.887 9	29.50	0.894 6	35.60	0.966 3
	CFSR［5］	38.07	0.960 7	33.74	0.919 2	32.24	0.900 5	32.28	0.930 0	39.00	0.977 8
	SAFMN［6］	38.00	0.960 5	33.54	0.917 7	32.16	0.899 5	31.84	0.925 6	38.71	0.977 1
	DSRNet［7］	37.61	0.958 4	33.30	0.914 5	31.96	0.896 5	31.41	0.920 9	-	-
	LapSRN［23］	37.52	0.959 1	32.99	0.912 4	31.80	0.895 2	30.41	0.910 3	37.27	0.974 0
	MemNet［24］	37.78	0.959 7	33.28	0.914 2	32.08	0.897 8	31.31	0.919 5	37.72	0.974 0
	CARN［22］	37.76	0.959 0	33.52	0.916 6	32.09	0.897 8	31.92	0.925 6	38.36	0.976 5
	IMDN［25］	38.00	0.960 5	33.63	0.917 7	32.19	0.899 6	32.17	0.928 3	38.88	0.977 4
	RLFN［26］	38.07	0.960 7	33.72	0.918 7	32.22	0.900 0	32.33	0.929 9	-	-
	ESRT［27］	-	-	-	-	-	-	-	-	-	-
	本文	38.07	0.960 1	33.89	0.919 9	32.23	0.899 7	32.76	0.942 3	39.05	0.977 2

Table 5. Comparison of PSNR/SSIM values of different algorithms under 3× reconstruction

View table

View in Article

Table 5. Comparison of PSNR/SSIM values of different algorithms under 3× reconstruction

尺度	方法	Set5		Set14		BSD100		Urban100		Manga109
尺度	方法	PSNR	SSIM	PSNR	SSIM	PSNR	SSIM	PSNR	SSIM	PSNR	SSIM
×3	Bicubic	30.39	0.868 2	27.55	0.774 2	27.21	0.738 5	24.46	0.734 9	26.95	0.855 6
	SRCNN［4］	32.75	0.909 0	29.30	0.821 5	28.41	0.786 3	26.24	0.798 9	30.48	0.911 7
	CFSR［5］	34.50	0.927 9	30.44	0.843 7	29.16	0.806 6	28.29	0.855 3	33.86	0.946 2
	SAFMN［6］	34.34	0.926 7	30.33	0.841 8	29.08	0.804 8	27.95	0.847 4	33.52	0.943 7
	DSRNet［7］	33.92	0.922 7	30.10	0.837 8	28.90	0.800 3	27.63	0.840 2	-	-
	LapSRN［23］	33.81	0.922 0	29.79	0.832 5	28.82	0.798 0	27.07	0.827 5	32.21	0.935 0
	MemNet［24］	34.09	0.924 8	30.00	0.835 0	28.96	0.800 1	27.56	0.837 6	32.51	0.936 9
	CARN［22］	34.29	0.925 5	30.29	0.840 7	29.06	0.803 4	28.06	0.849 3	33.50	0.944 0
	IMDN［25］	34.36	0.927 0	30.32	0.841 7	29.09	0.804 6	28.17	0.851 9	33.61	0.944 5
	RLFN［26］	-	-	-	-	-	-	-	-	-	-
	ESRT［27］	34.42	0.926 8	30.43	0.843 3	29.15	0.806 3	28.46	0.857 4	33.95	0.945 5
	本文	34.46	0.927 3	30.49	0.844 1	29.15	0.806 5	28.09	0.858 1	33.79	0.945 4

Table 6. Comparison of PSNR/SSIM values of different algorithms under 4× reconstruction

View table

View in Article

Table 6. Comparison of PSNR/SSIM values of different algorithms under 4× reconstruction

尺度	方法	Set5		Set14		BSD100		Urban100		Manga109
尺度	方法	PSNR	SSIM	PSNR	SSIM	PSNR	SSIM	PSNR	SSIM	PSNR	SSIM
×4	Bicubic	28.42	0.810 4	26.00	0.702 7	25.96	0.667 5	23.14	0.657 7	24.89	0.786 6
	SRCNN［4］	30.48	0.862 8	27.50	0.751 3	26.90	0.710 1	24.52	0.722 1	27.58	0.855 5
	CFSR［5］	32.33	0.896 4	28.73	0.784 2	27.63	0.738 1	26.21	0.789 7	30.72	0.911 1
	SAFMN［6］	32.18	0.894 8	28.60	0.781 3	27.58	0.735 9	25.97	0.780 9	30.43	0.906 3
	DSRNet［7］	31.71	0.887 4	28.38	0.773 7	27.43	0.730 3	25.65	0.769 3	-	-
	LapSRN［23］	31.54	0.885 2	28.09	0.770 0	27.32	0.727 5	25.21	0.756 2	29.09	0.890 0
	MemNet［24］	31.74	0.889 3	28.26	0.772 3	27.40	0.728 1	25.50	0.763 0	29.42	0.894 2
	CARN［22］	32.13	0.893 7	28.60	0.780 6	27.58	0.734 9	26.07	0.783 7	30.47	0.908 4
	IMDN［25］	32.21	0.894 8	28.58	0.781 1	27.56	0.735 3	26.04	0.783 8	30.45	0.907 5
	RLFN［26］	32.24	0.895 2	28.62	0.781 3	27.60	0.736 4	26.17	0.787 7	-	-
	ESRT［27］	32.19	0.894 7	28.69	0.783 3	27.69	0.737 9	26.39	0.796 2	30.75	0.910 0
	本文	32.46	0.897 3	28.79	0.785 6	27.69	0.740 4	25.85	0.782 7	31.01	0.914 4

Tools

Get Citation

Copy Citation Text

Jianpu LIN, Zhencheng WU, Kunfu WANG, Zhixian LIN, Tailiang GUO, Shanling LIN. Cascade residual-optimized image super-resolution reconstruction in Transformer network[J]. Optics and Precision Engineering, 2024, 32(12): 1902

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites

Paper Information

Category:

Received: Dec. 13, 2023

Accepted: --

Published Online: Aug. 28, 2024

The Author Email: Shanling LIN (sllin@fzu.edu.cn)

DOI:10.37188/OPE.20243212.1902

Topics

laser devices and laser physics

Lasers and Laser Optics

Laser physics

laser manufacturing

Instrumentation, Measurement and Metrology

Table 1. Simulation environment and parameter setting

Table 1. Simulation environment and parameter setting

Table 2. Validity verification of different modules

Table 2. Validity verification of different modules

Table 3. Network performance with different number of modules

Table 3. Network performance with different number of modules

Table 4. Comparison of PSNR/SSIM values of different algorithms under 2× reconstruction

Table 4. Comparison of PSNR/SSIM values of different algorithms under 2× reconstruction

Table 5. Comparison of PSNR/SSIM values of different algorithms under 3× reconstruction

Table 5. Comparison of PSNR/SSIM values of different algorithms under 3× reconstruction

Table 6. Comparison of PSNR/SSIM values of different algorithms under 4× reconstruction

Table 6. Comparison of PSNR/SSIM values of different algorithms under 4× reconstruction

微信扫一扫：分享