Object extraction from underwater images through logical stochastic resonance.

Area of Science:

• Image Processing
• Nonlinear Dynamics
• Signal Processing

Background:

• Underwater images suffer from heavy noise and nonlinearity due to suspended particles, complicating conventional processing.
• Inherent noise and nonlinearity in degraded images hinder accurate object extraction.

Purpose of the Study:

• To investigate the application of Logical stochastic resonance (LSR) for object extraction from highly degraded underwater images.
• To leverage the interplay of noise and nonlinearity for improved image analysis.

Main Methods:

• Degraded images are converted into a 1D form based on illumination.
• Normalized 1D image data is processed within the LSR system.
• Auxiliary Gaussian noise is introduced to enhance object-background separation.

Main Results:

• Demonstrated the effectiveness of LSR in processing natural offshore underwater images.
• Successfully improved the separation of objects from backgrounds in heavily degraded visuals.

Conclusions:

• LSR offers an effective approach for processing severely degraded underwater images.
• The proposed method presents a novel direction for underwater image analysis and object extraction.