Adversarial network for multi-input image restoration under strong turbulence

Abstract

Turbulence generated by random ups and downs in the refractive index of the atmosphere produces varying degrees of distortion and blurring of images in the camera. Traditional methods ignore the effect of strong turbulence on the image. This paper proposes a deep neural network to enhance image clarity under strong turbulence to handle this problem. This network is divided into two sub-networks, the generator and the discriminator, whose functions are to mitigate the effects of turbulence on the image and to determine the authenticity of the recovered image. After extensive experiments, it is proven that the present network plays a role in mitigating the image degradation problem caused by atmospheric turbulence.