A New SAR Image Segmentation Algorithm for the Detection of Target and Shadow Regions

Area of Science:

• Remote Sensing
• Image Processing
• Signal Processing

Background:

• Synthetic Aperture Radar (SAR) provides all-weather, all-time data acquisition capabilities.
• SAR imaging mechanisms inherently introduce speckle noise, complicating image segmentation.
• Accurate segmentation of target and shadow regions in SAR imagery is challenging due to noise.

Purpose of the Study:

• To develop a novel SAR image segmentation method robust to speckle noise.
• To enable simultaneous segmentation of target and shadow regions.
• To improve segmentation performance in low signal-to-clutter ratio (SCR) environments.

Main Methods:

• The proposed method utilizes wavelet decomposition for noise reduction.
• A constant false alarm rate (CFAR) algorithm is integrated for enhanced segmentation.
• The combined Wavelet Decomposition-Constant False Alarm Rate (WD-CFAR) algorithm is applied.

Main Results:

• The WD-CFAR algorithm demonstrates insensitivity to speckle noise in SAR images.
• Simultaneous segmentation of target and shadow regions was achieved.
• Effective segmentation was observed even with low signal-to-clutter ratios (SCR).

Conclusions:

• The proposed WD-CFAR method is effective and feasible for SAR image segmentation.
• The algorithm exhibits good general applicability across various SAR images.
• This approach offers a significant improvement for challenging SAR image analysis tasks.