Underwater Image Enhancement Using Color Compensation and Multi-Scale Fusion with Quality Evaluation Metrics

Images acquired underwater are often characterized by poor contrast, visibility, and color information due to the absorption and scattering of light within the underwater environment. This affects the overall performance of underwater images, particularly in underwater exploration and robot navigation. This paper proposes an improved underwater image enhancement approach using color compensation and multi-scale fusion techniques. Initially, the red channel and white balancing are performed to compensate for the color information loss. Two improved images are formed using the gamma correction and detail enhancement approaches. Weight maps are calculated to focus on the salient features of the underwater images using the Laplacian contrast, saliency, and saturation features. Finally, the images are fused using the multi-scale approach based on the Gaussian and Laplacian pyramid operations. Post-processing operations are performed to enhance the overall visibility of the underwater images using CLAHE, contrast stretching, and saturation enhancement techniques.