An optimized GAN method based on the Que-Attn and contrastive learning for underwater image enhancement

Area of Science:

• Computer Vision
• Image Processing
• Artificial Intelligence

Background:

• Underwater image restoration is crucial for exploring valuable underwater resources.
• Existing methods often rely on hand-crafted features, limiting their performance.
• Degraded underwater images present significant restoration challenges.

Purpose of the Study:

• To propose an effective unsupervised generative adversarial network (GAN) for underwater image restoration.
• To address the limitations of prior-based methods in handling complex underwater image degradation.

Main Methods:

• Developed an unsupervised GAN incorporating contrastive learning.
• Utilized PatchNCE loss to maximize mutual information between input and output.
• Designed a query attention (Que-Attn) module for feature importance selection.

Main Results:

• The proposed model demonstrates superior performance in underwater image restoration.
• Experiments show effective generalization across multiple benchmark datasets.
• Outperforms existing state-of-the-art methods in comparative analyses.

Conclusions:

• The unsupervised GAN with contrastive learning offers a powerful solution for underwater image restoration.
• The query attention module enhances the model's ability to focus on relevant features.
• This approach advances the field of underwater image processing and analysis.