Infrared and visible image fusion using improved generative adversarial networks

Li Min; Sijian Cao; Huaici Zhao; Pengfei Liu

doi:10.3788/IRLA20210291

Journals >Infrared and Laser Engineering >Volume 51 >Issue 4 >Page 20210291 > Article

Infrared and Laser Engineering
Vol. 51, Issue 4, 20210291 (2022)

Infrared and visible image fusion using improved generative adversarial networks

Li Min¹, Sijian Cao¹, Huaici Zhao^2、*, and Pengfei Liu²

Author Affiliations

¹School of Mechanical Engineering, Shenyang Jianzhu University, Shenyang 110168, China

²Key Laboratory of Optical-Electronics Information Processing, Shenyang Institute of Automation, Chinese Academy of Sciences, Shenyang 110169, China

show less

DOI: 10.3788/IRLA20210291 Cite this Article

Li Min, Sijian Cao, Huaici Zhao, Pengfei Liu. Infrared and visible image fusion using improved generative adversarial networks[J]. Infrared and Laser Engineering, 2022, 51(4): 20210291 Copy Citation Text

show less

Fig. 1. Overall framework of network structure

Download full size | View in the Article

Fig. 2. Generator network structure

Download full size | View in the Article

Fig. 3. Discriminator network structure

Download full size | View in the Article

Fig. 4. Comparative experimental results of the TNO dataset

Download full size | View in the Article

Fig. 5. Comparative experimental results of the RoadScene dataset. (a) Infrared image；(b) Visible image；(c) DRTV；(d) CNN；(e) FusionGAN；(f) DIDFuse；(g) DDcGAN；(h) Proposed method

Download full size | View in the Article

Fig. 6. Loss function curve

Download full size | View in the Article

Dataset	Methods	AG	SF	${Q^{AB/F}}$	${Q_{CB}}$
TNO	DRTV	3.761	9.639	0.319	0.411
	CNN	4.700	11.489	0.332	0.463
	FusionGAN	4.014	10.006	0.313	0.425
	DIDFuse	4.644	11.771	0.395	0.472
	DDcGAN	5.529	13.044	0.356	0.456
	Proposed method	5.861	14.454	0.401	0.504
Road scene	DRTV	3.221	8.696	0.368	0.384
	CNN	4.484	10.536	0.398	0.384
	FusionGAN	3.290	8.426	0.278	0.387
	DIDFuse	5.253	14.149	0.469	0.452
	DDcGAN	5.200	13.580	0.423	0.461
	Proposed method	5.855	15.243	0.480	0.494

Table 1. Objective evaluation results of two comparison experiment

View in the Article

Methods	AG	SF	${Q^{AB/F}}$	${Q_{CB}}$
DDcGAN	5.529	13.044	0.356	0.456
DDcGAN+SN	5.670	13.062	0.368	0.469
DDcGAN+GSFB	5.746	13.841	0.388	0.484
DDcGAN+GSFB+SN	5.861	14.454	0.401	0.504

Table 2. Evaluation results of ablation experiment

Li Min, Sijian Cao, Huaici Zhao, Pengfei Liu. Infrared and visible image fusion using improved generative adversarial networks[J]. Infrared and Laser Engineering, 2022, 51(4): 20210291

Download Citation

Tools

Save the article for my favorites

Paper Information