Use Deep Learning to Restore Landsat 7 Gap Images

I am recently working on fixing the Landsat 7 gaps by using deep learning technique. Because I found that the tools and methods provided by USGS are not very ideal for me.

Depending on your scenarios, USGS give out two solutions:

for display only

for scientific analysis

My idea is simple: use the SLC-on Landsat 7 images and Landsat 5 as training datasets to train a deep neural network (a latest one, maybe ResNet or UNet for image fusion). Use the trained network to fill the gaps in the SLC-off images.

As one of the most accurately corrected remote sensing satellite, the images of Landsat 7 have a lot to discover. But the gaps since 2003 are annoying and have been the major obstacle to let people full utilize it. Hope this method could revive Landsat 7 images in better exploring the Earth dynamics.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s