Dual-Attention-Based Feature Aggregation Network for Infrared and Visible Image Fusion

Infrared and visible image fusion aims to produce fused images which retain rich texture details and high pixel intensity in the source images. In this article, we propose a dual-attention-based feature aggregation network for infrared and visible image fusion. Specifically, we first design a multibranch channel-attention-based feature aggregation block (MBCA) by generating multiple branches to suppress useless features from different aspects. This block is also able to adaptively aggregate the meaningful features by exploiting the interdependencies between channel features. To gather more meaningful features during the fusion process, we further design a global-local spatial-attention-based feature aggregation block (GLSA), for progressively integrating features of source images. After that, we introduce multiscale structural similarity (MS-SSIM) as loss function to evaluate the structural differences between the fused image and the source images from multiple scales. In addition, the proposed network involves strong generalization ability since our fusion model is trained on the RoadScene dataset and tested directly on the TNO and MSRS datasets. Extensive experiments on these datasets demonstrate the superiority of our network compared with the current state-of-the-art methods. The source code will be released at https://github.com/tangjunyang/Dualattention.

Automated summary

In this article, a dual-attention-based feature aggregation network is proposed for infrared and visible image fusion. The network effectively aggregates useful features and adaptively integrates meaningful features through multiple branch channel attention and global-local spatial attention. The fusion process is evaluated using multiscale structural similarity as a loss function. Extensive experiments demonstrate the superiority of the proposed network compared to state-of-the-art methods on multiple datasets.