Gradient-based contrast enhancement and color correction for underwater images

Abstract

Underwater images are easily degraded by various characteristics of underwater environment. In this paper, we propose an underwater image enhancement algorithm which improves the low contrast and alleviates the color distortion in gradient domain. We first estimate the gradient of original scene image by scaling the gradient of an input underwater image with a corresponding transmission map. Then we reconstruct an original scene image from the compensated gradient image using the constraint that the three color components at each boundary pixel of the original image have the same value to the median value among the three color components of the input underwater image. Experimental results demonstrate that the proposed algorithm improves the contrast and alleviates the color distortion faithfully, and outperforms the state-of-the-art underwater image enhancement method.