DNDT: Infrared and Visible Image Fusion Via DenseNet and Dual-Transformer
Haibo Zhao, Rencan Nie
Abstract
Image fusion plays an important role in real life, especially in remote sensing, image enhancement, and so on. Among all kinds of image fusion algorithms, Transformer has been proposed recently for image fusion with great potential, but it also limited localization abilities due to insufficient low-level details. To address this issue, we proposed a new fusion framework called DenseNet Dual-Transformer(DT) for infrared and visible image fusion. It extracts rich feature information through the encoder of DenseNet, On the other hand, utilized DT to pay attention to different aspects of information, and integrate all aspects of information. We argue that DT as a fusion strategy, local and remote information can be capture. A large number of experiments have proved that the performance of the fusion algorithm proposed in the paper is better than many existing algorithms.