Automatic underwater polarization imaging without background region or any prior

Area of Science:

• Optical Engineering
• Image Processing
• Remote Sensing

Background:

• Traditional polarization underwater imaging relies on physical scattering models, often requiring background regions and prior knowledge, limiting practical applications.
• Existing methods struggle with non-uniform illumination and inaccurate background identification, leading to unstable image recovery performance.
• Variations in recovery results arise from subjective background selection and inconsistent parameter adjustments in conventional techniques.

Purpose of the Study:

• To develop an automatic underwater image recovery method that does not require background regions or prior knowledge.
• To optimize the physically feasible region within the underwater imaging model.
• To improve the robustness and adaptability of underwater image restoration.

Main Methods:

• Analysis and optimization of the physically feasible region in underwater imaging.
• Degeneration of intermediate variables in the physical scattering model.
• Development of a method adaptable to non-uniform illumination and difficult-to-identify background regions.

Main Results:

• Achieved automatic underwater image recovery without reliance on background regions or prior information.
• Demonstrated adaptability to underwater images with non-uniform illumination.
• Successfully enhanced image contrast while preserving details and minimizing noise in various underwater scenes.
• Validated effectiveness in dense, turbid water conditions.

Conclusions:

• The proposed method offers a practical and robust solution for automatic underwater image recovery.
• It overcomes limitations of traditional methods by eliminating the need for background regions and prior knowledge.
• The technique provides stable and consistent performance across diverse underwater imaging scenarios, including challenging turbid environments.