Fig. 1. Overall framework of the proposed method

Fig. 2. Generator network architecture

Fig. 3. Discriminator network architecture

Fig. 4. Effect of loss function on results. (a) Adversarial loss+content loss; (b) adversarial loss+perceptual loss; (c) adversarial loss+content loss+perceptual loss

Fig. 5. Influence of network structure on results. (a) Low-level image spatial combination information result; (b) high-level feature space combined information results

Fig. 6. Fusion results. (a) Low resolution infrared long wave images; (b) low-resolution infrared short wave images; (c) low resolution visible images; (d) CNN; (e) GAN; (f) Densefuse; (g) DGCNN; (h) proposed method

Fig. 7. Objective evaluation results.(a) Evaluation metrics for the first fused results; (b) evaluation metrics for the second fused results; (c) evaluation metrics for the third fused results; (d) evaluation metrics for the fourth fused results; (e) evaluation metrics for the fifth fused results; (f) evaluation metrics for the sixth fused results

Fig. 8. Fusion results. (a) Low resolution infrared long wave image; (b) low-resolution infrared short wave image; (c) low resolution visible image; (d) CNN; (e) GAN; (f) Densefuse; (g) DGCNN; (h) proposed method

Table 1. Objective evaluation index values