Fast and Efficient Zero-Learning Image Fusion

This repository provides an opencv and pytorch implementation of the paper "Fast and Efficient Zero-Learning Image Fusion".

Usage

pip install -r requirements.txt
cd code/
# --imageSource specify a sequence of source images to be fused
python main.py --imagePath=../images/IV_images --imageSource "VIS*.png" "IR*.png"
python main.py --imagePath=../images/MRI-PET --imageSource "MRI*.png" "PET*.png"

In the above example, VIS01.png will be fused with IR01.png, VIS02.png will be fused with IR02.png, etc.

Fusion of 1 RGB with multiple IR images are supported, just add the glob pattern of images in --imageSource.

Results

Visible and infrared Image fusion

Grayscale and RGB image fusion

License

This project is licensed under the MIT License.

MIT © bsun