A LIGHTWEIGHT IMAGE FUSION METHOD FOR INFRARED AND VISIBLE IMAGES

Infrared and visible image fusion integrates the complementary information from both modalities, generating more informative and richer images.With the increasing demands in security monitoring, autonomous driving and remote sensing, studies on lightweight infrared and visible fusion has become highly significant.Lightweight fusion models can reduce computational resource consumption, improve algorithmic processing speed, and make the model more suitable for hardware environments with limited resources.In this case, this paper proposes a lightweight infrared and visible image fusion method to enhance the fusion speed and broad adaptability.A systematic analysis of the fusion strategies for lightweight models is conducted.Extensive experiments are carried out on four mainstream infrared and visible fusion datasets.The experimental results show that, compared to mainstream fusion algorithms, the proposed lightweight fusion method achieves the best values in terms of peak signal-to-noise ratio (PSNR), mean squared error (MSE), structural similarity index measure (SSIM) and correlation coefficient (CC).The weight of the proposed fusion model is only 0.17M, and its fusion speed on the same hardware outperforms that of contrast fusion methods.This method provides effective support for the application of lightweight image fusion techniques on resource-constrained platforms.