Underwater Image Enhancement using Dehazing and Fusion Techniques

Abstract

Acquiring clear images in underwater environments is an important issue inocean engineering. The quality of underwater images plays a pivotal role in scientific missions such as monitoring sea life, taking census of populations, and assessing geological or biological environments. Capturing images underwater is challenging, mostly due to haze caused by light that is reflected from a surface and is deflected and scattered by water particles, and colour change due to varying degrees of light attenuation for different wavelengths. Light scattering and colour change result in contrast loss and colour deviation in images acquired underwater. The well-established underwater image fusion approach has been enhanced as a consequence of our study. Our underwater image improvement system consists of three primary components: slicing the original image into illuminance and reflectance slices, applying a linear piecewise colour correction algorithm to the reflectance slice, and contract enhancing algorithms to the illuminance slice. In order to recreate the final image, we use a multi-stage fusion approach based on Gaussian and Laplacian pyramids. Existing fusion-based approaches have been enhanced, as shown by experimental data