Ship Segmentation in SAR Images by Improved Nonlocal Active Contour Model

Area of Science:

• Remote Sensing
• Image Processing
• Oceanography

Background:

• Synthetic Aperture Radar (SAR) is crucial for ocean surveillance and ship monitoring.
• Ship segmentation in SAR images is challenging due to speckle noise and complex backscattering.
• Accurate ship segmentation is vital for shipping management and military applications.

Purpose of the Study:

• To propose a new nonlocal processing method for robust ship segmentation in SAR images.
• To address challenges posed by speckle noise and intensity variations.
• To achieve refined segmentation of ship targets in diverse SAR data.

Main Methods:

• Developed a novel nonlocal energy function for patch comparison and region optimization.
• Introduced a ratio distance metric to mitigate multiplicative noise effects.
• Incorporated integral histograms to accelerate convergence in pairwise interactions.

Main Results:

• The proposed method demonstrates robustness against speckle noise and intensity variations.
• Experimental results on real SAR data show improved segmentation accuracy.
• The technique is effective across different SAR resolutions and bands.

Conclusions:

• The nonlocal processing approach offers a significant advancement in SAR ship segmentation.
• This method enhances the reliability of ship monitoring for maritime applications.
• The findings contribute to more effective SAR image interpretation for ocean surveillance.