Color Balance and Fusion for Underwater Image Enhancement

Abstract

We introduce an effective technique to enhance the 
images captured underwater and degraded due to 
the medium scattering and absorption. Our method 
is a single image approach that does not require 
specialized hardware or knowledge about the 
underwater conditions or scene structure. It builds 
on the blending of two images that are directly 
derived from a color compensated and white
balanced version of the original degraded image. 
The two images to fusion, as well as their associated 
weight maps, are defined to promote the transfer of 
edges and color contrast to the output image. To 
avoid that the sharp weight map transitions create 
artifacts in the low frequency components of the 
reconstructed image, we also adapt a multiscale 
fusion strategy. Our extensive qualitative and 
quantitative evaluation reveals that our enhanced 
images and videos are characterized by better 
exposedness of the dark regions, improved global 
contrast, and edges sharpness. Our validation also 
proves that our algorithm is reasonably independent 
of the camera settings, and improves the accuracy of 
several image processing applications, such as 
image segmentation and key point matching. Index 
Terms— Underwater, image fusion, white
balancing.