LENFusion

LENFusion: A Joint Low-Light Enhancement and Fusion Network for Nighttime Infrared and Visible Image Fusion

This is official code of "LENFusion: A Joint Low-Light Enhancement and Fusion Network for Nighttime Infrared and Visible Image Fusion".

Before Train

conda env create -f LENFusion.yaml

Train

During training, eval is used for evaluation, and model is used to store the training weight.

The code has two training datasets in the train folder:

1. LFN_traingdata is the training dataset of Luminance Feedback Network (LFN). Links: google drive and baidu.
2. vi and ir are the training dataset of the main networks (Luminance Adjustment Network and Re-enhancement Fusion Network, LAN and RFN). Links: google drive and baidu.

Test

Download checkpoint at the links: google drive and baidu. The test data from the KAIST, LLVIP, MSRS, TNO datasets.

The Environment

numpy=1.26.0

opencv-python=4.8.1.78

python=3.10.13

torch=1.12.1

torchvision=0.13.1

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

@article{Chen2024LENFusion,
  title={LENFusion: A Joint Low-Light Enhancement and Fusion Network for Nighttime Infrared and Visible Image Fusion},
  author={Chen, Jun and Yang, Liling and Liu, Wei and Tian, Xin and Ma, Jiayi},
  journal={IEEE Transactions on Instrumentation and Measurement},
  volume={},
  pages={},
  year={2024}
}