A PolSAR Image Segmentation Algorithm Based on Scattering Characteristics and the Revised Wishart Distance.

Area of Science:

• Remote Sensing
• Image Processing
• Geospatial Analysis

Background:

• Polarimetric Synthetic Aperture Radar (PolSAR) imaging provides rich scattering information.
• Accurate segmentation of PolSAR images is crucial for various applications.
• Existing segmentation methods often struggle with speckle noise and detail preservation.

Purpose of the Study:

• To propose a novel segmentation algorithm for PolSAR images.
• To improve speckle suppression and detail preservation in PolSAR image segmentation.
• To achieve more accurate segmentation results compared to existing methods.

Main Methods:

• A two-component algorithm: merging order and merging predicate.
• Utilizes complex-kind Hotelling-Lawley trace (HLT) statistic for merging order.
• Employs scattering characteristics and revised Wishart distance for merging predicate.
• Includes iterative decomposition and merging with a postprocessing step.

Main Results:

• Demonstrated superior performance in speckle suppression.
• Showcased enhanced detail preservation capabilities.
• Achieved more accurate segmentation results on RADARSAT-2 and Gaofen-3 PolSAR images.

Conclusions:

• The proposed algorithm offers significant improvements in PolSAR image segmentation.
• It effectively addresses challenges of speckle noise and detail loss.
• The method provides a robust solution for accurate PolSAR data analysis.