Adaptive Ship Detection for Single-Look Complex SAR Images Based on SVWIE-Noncircularity Decomposition

Area of Science:

• Remote Sensing
• Signal Processing
• Radar Imaging

Background:

• Synthetic Aperture Radar (SAR) imaging is crucial for maritime surveillance.
• Accurate ship detection in complex backgrounds remains a challenge due to clutter and noise.
• Existing methods often struggle with targets near islands or in high-noise environments.

Purpose of the Study:

• To develop an adaptive ship detection method for single-look complex SAR images.
• To enhance the performance of ship detection in challenging marine scenarios.
• To improve the accuracy and robustness of automated target recognition in SAR data.

Main Methods:

• Analysis and application of noncircularity for ship detection.
• Introduction of Similarity Variance Weighted Information Entropy (SVWIE) for clutter reduction and target enhancement.
• Development of SVWIE-noncircularity (SN) decomposition yielding high-noncircularity (h) and low-noncircularity (l) components.
• Utilizing the generalized Gamma distribution (GΓD) to model clutter statistics of the 'h' component.
• Employing Constant False Alarm Rate (CFAR) for adaptive threshold selection.

Main Results:

• Demonstration that the 'h' component effectively differentiates ships from clutter in SAR images.
• Successful ship detection in complex backgrounds, including scenarios with ships near small islands or strong noise.
• Validation of the proposed method using HH polarization Alos-2 SAR images.

Conclusions:

• The proposed adaptive ship detection method offers improved accuracy and robustness for SAR imagery.
• The SN decomposition and CFAR-based thresholding effectively address challenges in complex maritime environments.
• This technique shows significant potential for enhanced maritime surveillance and target recognition using SAR data.