Underwater Image Enhancement via L2 based Laplacian Pyramid Fusion

Abstract

Underwater images are known to suffer from color shift and poor visibility because light is absorbed and scattered as it travels through the water. In this paper, we present an efficient fusion-based framework to enhance underwater images. Our approach has three main steps. First, color correction is used to eliminate color cast and preliminary results were obtained. Then, we processed the lumimance channel through enhancing contrast, and performe Laplacian pyramid decomposition for it and the lumimance channel after color correction respectively. Finally, we use proposed fusion rules, optimal balance based on L2 norm, to merge layer by layer the two pyramids obtained in the second step separately. Experiments show the effectiveness of the proposed method than others in restoring the original color and enhancing underwater image contrast and visibility.