Advanced Underwater Imagery Enhancement using CycleGAN for Marine Research and Photography

This paper proposes the Underwater Image Enhancement using CycleGAN generative adversarial network (GAN). It can be seen that underwater images encounter kinds of color distortion, low contrast, and haziness due to the interaction of light with water. CycleGAN helps our system to overcome the challenge of fine tuning raw underwater images to better quality images in terms of color and contrast. The goal of this method is to serve those in marine biology, underwater photography, and research by delivering superior visual data for enhanced identification and representation purposes. We mention below that we have compared the outcome with existing approaches, judged by standard image quality measures. Consequently, the results show the possible application of CycleGAN for achieving better image quality with minimized preprocessing time.