Underwater image sharpening based on structure restoration and texture enhancement

Area of Science:

• Computer Vision
• Image Processing

Background:

• Underwater optical images suffer from blurring and color distortion due to light absorption and scattering.
• Existing methods struggle to effectively restore visual quality in underwater environments.

Purpose of the Study:

• To propose a novel image sharpening method for improving the visual quality of underwater optical images.
• To address issues of blurring, color distortion, and atomization in underwater imagery.

Main Methods:

• Image decomposition into structure and texture layers using a relative total variation model.
• Application of the red-blue dark channel prior (RBDCP) model for background light calculation and transmission map generation.
• Detail enhancement using a combination of Gaussian kernel and binary mask, followed by layer fusion for sharpening.

Main Results:

• The RBDCP model effectively calculates background light and corrects color through red-blue channel analysis.
• The detail lifting algorithm enhances texture information.
• The fused image exhibits improved hue, saturation, clarity, and reduced atomization.

Conclusions:

• The proposed fusion-based image sharpening method effectively restores underwater image quality.
• The technique demonstrates superior performance in color correction, clarity enhancement, and atomization removal compared to existing methods.