Abstract: In order to solve the problems of low contrast and unclear detail information of composite CT images under single X-ray energy, a dual-energy CT image fusion method based on generative adversarial network was proposed to fuse images under low and high energy condition. The network consists of a generator and two discriminators. The generator was used to extract the details of CT images. Meanwhile, two discriminators were used to distinguish the structural differences between fusion CT images and high-low energy CT images, and to distinguish the authenticity of data. The fusion model was established through the adversarial training of the end to end model, and finally the fusion CT image containing high and low energy information was generated. The experimental results show that the proposed dual-energy image fusion algorithm based on generative adversarial network solves the limitations of single-energy CT imaging well, and the fused dual-energy CT image has more details, which was beneficial to the interpretation of key image information of composite materials.