Deep Submerged Image Enhancement and Restoration Process using CNN

Abstract

In oceanographic studies, underwater imagery plays a vital role. Underwater imaging has some of the advanced applications such as hand-held stereo-cam, fish-pond monitoring, etc. The major sources of quality degradation in most of the underwater imaging processes are scattering and absorption which occurs due to light assimilation. In this paper, we propose a two step-strategy in which the former is the enhancement process and latter is the restoration process. Our unavoidable selective and quantitative appraise uncover that our upgraded pictures and recordings have better accessibility in the dark locales, progressed global and local contrast and better edge sharpness. In order to get rid of image quality impairments, we follow a method which involves only a single image. The major advantage of this method is that it does not require a specialized image-capturing equipment. Moreover, our substantiation gives a better accuracy by deploying Convolutional Neural Network(CNN) algorithm.