Multi-Feature Fusion Network for Infrared and Visible Image Fusion

Infrared and visible image fusion aims to integrate the complementary information in the two types of images to generate fusion images. In the current infrared and visible image fusion tasks, effectively integrating features at different levels remains a challenge. For this purpose, we propose a new multifeature fusion network, MFFNet. This network builds multilayer feature extraction modules, starting from extracting shallow features to avoid losing important detail information. In addition, MFFNet fuses multi-level infrared and visible features through a novel attention-guided fusion module. Specifically, the module refines the extracted features by extracting multi-scale spatial information, enabling it to focus more on the targets in the infrared image and the visible details, while preserving the complementary information of the two modalities. Finally, the image reconstruction layer integrates multiple fused features to generate the fused image. The experimental results indicate that the proposed fusion method demonstrates superior performance compared with existing fusion methods.