<p>Currently, many image fusion methods can achieve excellent performance. However, the use of cross-modal image information can still be explored further. Most methods utilize the powerful feature representation capabilities of deep neural networks to fuse image features at a deep level. However, this ignores the correlation between the original image information. To better utilize the global information and cross-modal correlation information of the original image itself, this paper proposes an infrared and visible light image fusion algorithm based on cross-modal interaction and gating mechanism. First, features are extracted from the source image through a two-branch network. After cross-modal feature interaction, the initial fused features and the original image features are then fused again through the feature selection module. The secondary fusion process can better utilize the cross-modal correlation of the original information. Finally, the fusion decoding module gradually completes the guided fusion from deep to shallow layers. Experimental results show that the method proposed in this paper has significant improvements compared with existing advanced technologies.</p>