Turbidity Underwater Image Enhancement based on Generative Adversarial Network

In recent years, with the continuous development of underwater object detection and recognition applications, the requirements for underwater image quality are getting higher and higher. However the quality of underwater image is always degraded seriously due to the light absorption, scattering of water itself and the suspended particles in water. As a result, the underwater images always suffer from noise pollution, reduced contrast, color distortion and blurred texture, etc. With the development of neural networks, they are applied to solve the problem of underwater enhancement as well. Due to the limited learning ability of classical deep networks, the distortion of underwater images cannot be removed thoroughly. Therefore a generative adversarial network is proposed in this paper for underwater image enhancement. In the experiments, the underwater images of different states were widely tested, and the proposed generative adversarial network improved the image quality and texture details better compared with the traditional enhancement method and classical residual networks.