yoghourtcover/ddcgan

Code of DDcGAN for infrared and visible image fusion and medical image fusion

DDcGAN-tensorflow:
infrared and visible image fusion via dual-discriminator conditional generative adversarial network

This work can be applied for

1. multi-resolution infrard and visible image fusion
   
2. same-resolution infrared and visible image fusion
   
3. PET and MRI image fusion
   

Framework:

Generator architecture:

Training dataset:

1. vis-ir dataset (password:nh2r).
   
2. PET-MRI dataset (password: 5d9y).
   The code to create your own training dataset can be found here.

If this work is helpful to you, please cite it as:

@article{ma2020ddcgan,
  title={DDcGAN: A Dual-Discriminator Conditional Generative Adversarial Network for Multi-Resolution Image Fusion},
  author={Ma, Jiayi and Xu, Han and Jiang, Junjun and Mei, Xiaoguang and Zhang, Xiao-Ping},
  journal={IEEE Transactions on Image Processing},
  volume={29},
  pages={4980--4995},
  year={2020},
  publisher={IEEE}
}s

The previous version of our work can be seen in this paper:

@inproceedings{xu2019learning,
  title={Learning a generative model for fusing infrared and visible images via conditional generative adversarial network with dual discriminators},
  author={Xu, Han and Liang, Pengwei and Yu, Wei and Jiang, Junjun and Ma, Jiayi},
  booktitle={proceedings of Twenty-Eighth International Joint Conference on Artificial Intelligence (IJCAI-19)},
  pages={3954--3960},
  year={2019}
}

This code is base on the code of DenseFuse.

If you have any question, please email to me ([email protected]).