Underwater image enhancement using multi-task fusion.

Area of Science:

• Computer Vision
• Image Processing
• Oceanography

Background:

• Underwater images suffer from scattering, absorption, and color distortion due to water conditions.
• These impairments reduce visibility and contrast, hindering advanced visual analysis.
• Existing methods often fail to address multiple degradation factors simultaneously.

Purpose of the Study:

• To propose a novel underwater image enhancement method based on multi-task fusion (MTF).
• To address color distortion, low visibility, and poor contrast in underwater imagery.
• To achieve superior image restoration and enhancement for subsequent visual tasks.

Main Methods:

• Color correction using linear constraints and the gray world assumption.
• Visibility enhancement via an improved type-II fuzzy set-based algorithm.
• Contrast enhancement using standard normal distribution and softplus function.
• Multi-task fusion by solving for similarity and deriving optimal fusion weights.

Main Results:

• The proposed MTF method demonstrates excellent image enhancement and restoration capabilities.
• MTF produces visually pleasing results with improved clarity and color accuracy.
• Quantitative and qualitative evaluations show MTF outperforms ten state-of-the-art algorithms on two datasets.

Conclusions:

• The multi-task fusion approach effectively enhances underwater image quality.
• MTF shows significant improvements in image restoration and visual appeal.
• The method shows promise for applications like target and edge detection in underwater environments.