Lightweight infrared image super-resolution reconstruction based on gradient guidance

Kun ZHU; Liangji SHEN; Wentao JIANG; Chaojie YE; Zhonghao LI; Wei WEI; Jilong LIU

doi:10.5768/JAO202546.0304002

Journal of Applied Optics, Volume. 46, Issue 3, 695(2025)

Lightweight infrared image super-resolution reconstruction based on gradient guidance

Kun ZHU, Liangji SHEN^*, Wentao JIANG, Chaojie YE, Zhonghao LI, Wei WEI, and Jilong LIU

Author Affiliations

Xi'an Institute of Applied Optics, Xi'an 710065, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(12)

Fig. 1. Structure diagram of infrared image super-resolution reconstruction network

Download full size

View in Article

Fig. 2. Part of images from dataset 2

Download full size

View in Article

Fig. 3. Comparison of results for different super-resolution methods under scenario 1 (with scale factor of 2)

Download full size

View in Article

Fig. 4. Comparison of results for different super-resolution methods under scenario 2 (with scale factor of 4)

Download full size

View in Article

Fig. 5. Comparison of results for different super-resolution methods under scenario 3 (with scale factor of 2)

Download full size

View in Article

Fig. 6. Comparison of results for different super-resolution methods under scenario 4 (with scale factor of 4)

Download full size

View in Article

Fig. 7. Comparison of results for different super-resolution methods under scenario 5 (with scale factor of 2)

Download full size

View in Article

Fig. 8. Comparison of results for different super-resolution methods under scenario 6 (with scale factor of 4)

Download full size

View in Article

Fig. 9. Comparison of $M_{1}$ model and $M_{2}$ results under the same scenario

Download full size

View in Article

Table 1. Results of comparative experiment (dataset 1)

View table

View in Article

Table 1. Results of comparative experiment (dataset 1)

Method	Scale	PSNR	SSIM	Params/K	FLOPs/G
SRCNN	×2	30.23	0.8243	57.3	17.57
FSRCNN		34.62	0.8734	12.8	2.02
IMDN		35.00	0.8948	694.4	52.95
MSID		34.10	0.8648	375.5	27.34
BSRN		35.56	0.9096	331.6	24.02
Omni-SR		35.65	0.9082	771.7	57.48
SAFMN		37.19	0.9373	226.7	17.16
YUAN		38.96	0.9627	81.1	6.25
SHI		39.46	0.9675	255.1	21.83
Ours		39.76	0.9699	111.4	8.54
SRCNN	×4	29.73	0.7086	57.3	17.57
FSRCNN		31.10	0.7336	12.8	1.55
IMDN		33.07	0.7937	715.2	13.64
MSID		31.73	0.6834	393.6	7.18
BSRN		33.16	0.8086	352.4	6.40
Omni-SR		33.26	0.8043	792.5	14.77
SAFMN		33.42	0.7864	238.4	4.51
YUAN		35.62	0.8974	81.1	6.25
SHI		35.90	0.9047	255.1	30.92
Ours		35.91	0.9051	111.3	8.54

Table 2. Results of comparative experiment (dataset 2)

View table

View in Article

Table 2. Results of comparative experiment (dataset 2)

Methods	Scale	PSNR	SSIM	Params/K	FLOPs/G
BSRN	×2	33.65	0.7640	331.6	24.02
SAFMN		34.61	0.8301	226.7	17.16
YUAN		35.18	0.8562	81.1	6.25
SHI		35.57	0.8725	255.1	21.83
Ours		35.45	0.8726	111.4	8.54
BSRN	×4	33.95	0.7675	352.4	6.40
SAFMN		32.99	0.7182	238.4	4.51
YUAN		33.96	0.7714	81.1	6.25
SHI		34.03	0.7813	255.1	30.92
Ours		34.05	0.7762	111.3	8.54

Table 3. Comparison of objective evaluation metrics for various models
View table
View in Article
Table 3. Comparison of objective evaluation metrics for various models
Methods Scale GMG RB $α$ PSNR SSIM
$M_{1}$ ×2 5 0.01 36.77 0.9325
$M_{2}$ √ 5 0.01 39.13 0.9655
$M_{3}$ √ 5 0.1 38.72 0.9619
$M_{4}$ √ 5 0.001 39.04 0.9637
$M_{5}$ √ 10 0.01 39.31 0.9665
$M_{6}$ √ 15 0.01 39.76 0.9699

Tools

Get Citation

Copy Citation Text

Kun ZHU, Liangji SHEN, Wentao JIANG, Chaojie YE, Zhonghao LI, Wei WEI, Jilong LIU. Lightweight infrared image super-resolution reconstruction based on gradient guidance[J]. Journal of Applied Optics, 2025, 46(3): 695

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites