MBHFuse: A multi- branch heterogeneous global and local infrared and visible image fusion with differential convolutional amplification features (JOLT 2024)

Introduction

This is official implementation of "MBHFuse: A Multi- branch Heterogeneous Global and Local Infrared and Visible Image Fusion with Differential Convolutional Amplification Features" with Pytorch.

Tips

The Trained Model is here, and the code is: MBHF.

Recommended Environment

• CUDA 11.1
• conda 4.11.0
• Python 3.7.16
• PyTorch 1.9.0
• timm 0.9.7
• tqdm 4.65.0
• pandas 1.3.5

Framework

Structure of GFE

Structure of LFE

Comparison with SOTA methods

Fusion results on TNO dataset

Fusion results on RoadScene dataset

Fusion results on MSRS dataset

Fusion results on M3FD dataset

Fusion results on LLVIP dataset

Quantitative fusion results

Citation

If you find this repository useful, please consider citing the following paper:

@article{sun2024JOLT,
  title={MBHFuse: A multi- branch heterogeneous global and local infrared and visible image fusion with differential convolutional amplification features},
  author={Sun, Yichen and Dong, Mingli and Yu, Mingxin and Zhu, Lianqing},
  journal={Optics and Laser Technology},
  year={2025},
  volume={181},
  part={A},
  pages={111666},
  doi={10.1016/j.optlastec.2024.111666}
  }  

If you have any questions, feel free to contact me (sunyichen@emails.bjut.edu.cn)

Acknowledgements

Parts of this code repository is based on the following works:

• https://github.com/linklist2/PIAFusion_pytorch
• https://github.com/Zhaozixiang1228/MMIF-CDDFuse