Here are some more images to give more context to the paper,
Figure A1 : Discriminator architecture used for the adversarial training of the implemented approach
Input SAR Optical GT CycleGAN Pix2Pix Ours Adv. Scr.
Figure A2 : Results examples for an image without normalization – SAR image is displayed as (VV, 1/2 (V + VH), HH) and Optical image as RGB