CBLA: Color-Balanced Locally Adjustable Underwater Image Enhancement

The exponentially growing requirement of underwater exploration to gather information and for scientific experiments has made it paramount to devise methods to ease the study of these terrains. However, underwater scenes suffer from light scattering and attenuation, which often leads to color-distorted, blurred, and underexposed images. Despite their poor visibility, these scenes carry a lot of information that needs to be unveiled for further investigation of the scenes. This work proposes a novel color-balanced locally adjustable (CBLA) method to enhance these underwater scenes locally, thus preserving the information contained. The method initiates with restoring the color of the captured scene, which is distorted due to various underwater conditions. This is carried out in RGB color space. Furthermore, the method uses the CIELAB color space to enhance the image’s contrast and restore its naturalness. The contrast is enhanced by processing the illumination component, while the a and b color channels help restore the captured scene’s naturalness. The proposed work is analyzed extensively using both qualitative and quantitative parameters. An extensive set of three datasets have been considered for a detailed and versatile evaluation. The source code for the proposed work can be accessed at
https://github.com/Severus-Matthew/CBLA
.