This repository implements pansharpening and multispectral images fusion algorithms based on the wavelet transform calculated by the à trous algorithm, using Python. The images used are downloaded from Google Earth Engine.
Pansharpening is applied to Landsat 8 raw images, and multispectral images fusion is performed over Landsat 8 surface reflectance images and Sentinel 2 surface reflectance images.
To execute the jupyter notebooks, an environment including the dependencies is offered. This environment can be found in image_fusion.yml. After importing the environment in conda. You can proceed with the execution of the notebooks.
There are 2 main execution flows:
-
Pansharpening: To perform pansharpening, download the Landsat 8 images using Download_Landsat8_Images.ipynb (A Google Earth Engine validated account is requiered), and then use Pansharpening.ipynb, to fuse the multispectral and the panchromatic image.
-
Multispectral images fusion: To perform multispectral images fusion, download the Landsat 8 and Sentinel 2 images using Download_Landsat8_and_Sentinel2_Images.ipynb, then use Resizing_Module.ipynb to resize the Landsat 8 image to the Sentinel 2 image pixel size. Finally, you can do the multispectral images fusion executing Multispectral_Images_Fusion.ipynb.