Multiscale Fusion for Underwater Image Enhancement

Abstract

In this paper, an enhancement scheme on underwater images is proposed. Underwater images undergo efforts like scattering and attenuation. Many techniques are proposed for underwater image and enhancement that largely depends on the features and characteristics of light. The characterization of light signal is complex and the underlying system that depends on light will lead to additional complexities. In the proposed scheme, image fusion is done by blending two images that are generated from a single underwater image. The first image is pre-processed by white balancing and Gamma correction. The second image is pre-processed by white balancing and sharpening. White balancing by colour channels gives better enhancement both specially and temporally. The white balancing followed by Gama correction and sharpening highlights a regular underwater image in all geo-spatial locations. The multiscale fusion of Gama corrected and sharpened image produces an enhanced underwater image. Simulations are carried on large number of underwater images and it was established that the proposed scheme out performs state-of-the-art techniques.