Infrared and Visible Image Fusion Method Based on Dual-Channel Generative Adversarial Network

Chunping Hou; Xiaocong Wang; Han Xia; Yang Yang

doi:10.3788/LOP202158.1410024

Journals >Laser & Optoelectronics Progress >Volume 58 >Issue 14 >Page 1410024 > Article

Laser & Optoelectronics Progress
Vol. 58, Issue 14, 1410024 (2021)

Infrared and Visible Image Fusion Method Based on Dual-Channel Generative Adversarial Network

Chunping Hou¹, Xiaocong Wang¹, Han Xia², and Yang Yang^1、*

Author Affiliations

¹School of Electrical and Information Engineering,Tianjin University, Tianjin 300072, China

²Tianjin International Engineering Institute, Tianjin University, Tianjin 300072, China

show less

DOI: 10.3788/LOP202158.1410024 Cite this Article Set citation alerts

Chunping Hou, Xiaocong Wang, Han Xia, Yang Yang. Infrared and Visible Image Fusion Method Based on Dual-Channel Generative Adversarial Network[J]. Laser & Optoelectronics Progress, 2021, 58(14): 1410024 Copy Citation Text

show less

Fig. 1. Network architecture of our method

Download full size

Fig. 2. GAN structure based on dual-channel

Download full size

Fig. 3. Structure of the pixel attention module

Download full size

Fig. 4. Structure of the discriminator network

Download full size

Fig. 5. Fusion results of different methods. (a)--(e) Image 1--image 5

Download full size

Fig. 6. Partial enlarged view of Fig. 5(d). (a) IR; (b) VI; (c) CVT; (d) MSVD; (e) Wavelet; (f) CSR; (g) DLF; (h) FusionGAN; (i) DenseFuse; (g) LatLRR; (k) PMGI; (l) our method

Download full size

Fig. 7. Results of ablation experiments.(a)--(c) Image 1--image 3

Download full size

Evaluation	EN	SD	CC	SSIM	Q_CV	Q_CB
CVT	6.583	27.029	0.608	0.712	370.72	0.508
MSVD	6.408	23.744	0.630	0.743	401.314	0.491
Wavelet	6.321	23.194	0.636	0.773	356.350	0.487
CSR	6.415	25.030	0.626	0.743	390.768	0.491
DLF	6.535	35.279	0.635	0.716	317.850	0.461
FusionGAN	6.336	25.030	0.517	0.647	627.313	0.421
DenseFuse	7.048	39.306	0.594	0.670	350.909	0.489
LatLRR	6.343	23.510	0.635	0.772	365.984	0.483
PMGI	6.964	34.012	0.628	0.706	419.241	0.522
Ours	6.836	32.638	0.615	0.779	626.444	0.525

Table 1. Evaluation indicators of different fusion methods

Evaluation	EN	SD	CC	SSIM	Q_CV	Q_CB
no_att&single_ch	6.784	28.231	0.601	0.834	377.794	0.470
no_att&double_ch	7.300	45.745	0.646	0.799	405.544	0.496
att&single_ch	6.476	25.454	0.610	0.654	1052.110	0.479
att&double_ch	6.792	28.480	0.654	0.864	688.366	0.513

Table 2. Image fusion results of different methods

Download Citation

Set citation alerts for the article

Tools

Set citation alerts for the article

Save the article for my favorites

Paper Information