This section introduced how to produce one synthetic remote sensing image (i) from multiple sources, and (ii) in making use directly of TensorFlow to train the model from Python code. We have applied a method which performs the gapfilling of a cloudy optical image, from multiple optical images and a synthetic aperture radar image.
The dataset used during this exercise consisted of three optical Sentinel-2 images, including one scene polluted by clouds, and one SAR Sentinel-1 image. After we carefully rejected the cloudy parts of the images in the patch selection, we proceeded to the patch extraction to construct two datasets: one for the training, the other for validation.
Once the patch extraction was done, we trained our ...
Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.
Read now
Unlock full access