Hybrid Underwater Image Enhancement via Dual Transmission Optimization and Transformer-Based Feature Fusion.

Area of Science:

• Computer Vision
• Image Processing
• Underwater Imaging

Background:

• Underwater environments present significant challenges for image restoration due to scattering, absorption, and color distortion.
• Existing methods often struggle to effectively address these complex degradations, limiting accurate underwater image analysis.

Purpose of the Study:

• To develop a robust hybrid approach for significant underwater image enhancement.
• To improve visibility, reduce scattering effects, and correct color distortion in underwater imagery.

Main Methods:

• A hybrid method combining dual transmission estimation with boundary constraints and local contrast.
• Adaptive ambient light estimation and color correction for robust distortion removal.
• Utilizing a U-Net Transformer (Uformer) for effective global and local feature extraction.

Main Results:

• The proposed method successfully estimates transmission maps, addressing visibility degradation.
• Robust correction of color distortion is achieved through adaptive ambient light estimation.
• The Uformer network effectively enhances image details and contextual information.

Conclusions:

• The hybrid approach demonstrates superior performance in underwater image enhancement compared to state-of-the-art methods.
• Quantitative evaluations using UIQM and UCIQE metrics validate the effectiveness of the proposed technique.