Labeling Dataset Based Colorization of SAR Images Using Cycle GAN
Sam-Young Lee, Daewon Chung · The Journal of Korean Institute of Electromagnetic Engineering and Science · 2022
After microwave radiation, a synthetic aperture radar (SAR) sensor detects electromagnetic wave energy scattered by the object to be observed with an antenna and realizes the shape of the ground object as an image through analysis. However, owing to sensor characteristics, color information that can be acquired from an optical image is absent. Therefore, colorization studies of SAR images were conducted using image translation techniques such as pix2pix and Cycle GAN.e In this study, colorization studies of SAR images were conducted using a Cycle GAN, a concept developed to enable color realization even when a paired dataset does not exist.