Dehazing algorithms for visibility improvement of degraded single underwater images

Abstract

Media and imageries are potent and effective means of communication. They offer a meaningful interpretation of concepts and data. Images help people to process visual information faster with improved comprehension than text. The earth has over 71% of the water expanse, of which more than 80% is not explored or even seen. Underwater Images (UIs) using Remotely Operated Vehicles (ROV), such as underwater robots and submarines with suitable cameras, are required to discover the underwater environment. The images are obtained for the motive of archaeological survey, an inspection of flora, fauna, oil wells and computer vision applications. Computer vision is a branch of artificial intelligence allowing computers to obtain eloquent information with digital images or videos. Based on the image interpretation, necessary actions or recommendations are implemented. Images thus obtained have to be preprocessed before subjecting to it any application for accurate results. This preprocessing is mainly categorized as Enhancement (EN) and Restoration (RN). The EN operations aim to improve contrast and reduce colour cast, saturation and noise, and is achieved by working on the pixel values for the change. The RN operations, however, follow an Image Formation Classical (IFC) that mathematically defines the parameters for image formation on the device. Recent trends show a combination of EN and RN as a Hybrid (HY), including the pros of both EN and RN for better performance and the same is ventured in this research. newline