AT-GAN: A generative adversarial network with attention and transition for infrared and visible image fusion

Yujing Rao, Dan Wu, Mina Han, Ting Wang, Yang Yang, Tao Lei, Chengjiang Zhou, Haicheng Bai, Lin Xing

Abstract: Highlights•Constraining image features by semantic transition and instance attention modules.•Learning the fusion equilibrium point by the customized structure evaluation module.•Keeping a faster fused image generation by a light network.