"Visualisation of the proposed FusionBooster when applied to the infrared and visilble image fusion approach task."

1 FusionBooster

This is the offical implementation for the paper titled "FusionBooster: A Unified Image Fusion Boosting Paradigm".

Paper

2 Environment

python 3.7.3
torch 1.9.0
scipy 1.2.0

3 Test (End to end, you provide two source images, we directly output the enhanced fused image)

Backbone: MUFusion (If you want to report this result, please mark our mehtod as FusionBooster-MU)

3.1 For the IVIF task:

python test_e2e_rgb.py

3.2 For the MEIF task (to do):

python xxx.py

3.3 For the MFIF task (to do):

python xxx.py

4 Test (Booster Only, you provide the initial fused image, we enhance it)

To use our pre-trained FusionBooster to boost an arbitary method:

4.1 For the IVIF task (Backbone: DDcGAN)

python test_booster_only_rgb.py

4.2 For the MEIF task (to do)

python xxx.py

4.3 For the MFIF task (to do)

python xxx.py

You can modify the path in the "test_booster_only_xxxx.py" file, to enhance your own fusion results.

5 Train

4.1 For the IVIF task (Backbone: DDcGAN)

Training Set (DDcGAN Results on LLVIP) Password: hokf

Training Set (Original LLVIP) Password: jq15

5 Announcement

• 2024-10-14 The code for end-to-end boosting source images (IVIF) is now available. ("test_e2e_rgb.py").
• 2024-10-14 The code for boosting an arbitary method is available ("test_booster_only.py").
• 2024-10-1 Because some of the fusion methods are realised using the tensorflow framework. Our FusionBooster demo will be implemented based on the MUFusion. You can always use our "detached booster" to enhance your own fusion results.
• 2024-9-30 This work has been accepted by IJCV.

6 Highlight

• We devise an image fusion booster by analysing the quality of the initial fusion results by means of a dedicated Information Probe.
• The proposed FusionBooster is a general enhancer, which can be applied to various image fusion methods, e.g., traditional or learning-based algorithms, irrespective of the type of fusion task.
• In a new divide-and-conquer image fusion paradigm, the results of the analysis performed by the Information Probe guide the refinement of the fused image.
• The proposed FusionBooster significantly enhances the performance of the SOTA fusion methods and downstream detection tasks, with only a slight increase in the computational overhead.

7 Citation

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

@article{cheng2024fusionbooster,
  title={FusionBooster: A Unified Image Fusion Boosting Paradigm},
  author={Cheng, Chunyang and Xu, Tianyang and Wu, Xiao-Jun and Li, Hui and Li, Xi and Kittler, Josef},
  journal={International Journal of Computer Vision},
  year={2025}
}