Multilook SAR Image Segmentation with an Unknown Number of Clusters Using a Gamma Mixture Model and Hierarchical...

Area of Science:

• Remote Sensing
• Image Processing
• Computer Vision

Background:

• Multilook Synthetic Aperture Radar (SAR) imagery presents challenges in segmentation due to complex statistical properties.
• Accurate segmentation requires determining the unknown number of homogeneous regions within SAR images.

Purpose of the Study:

• To develop a novel algorithm for multilook SAR image segmentation that automatically determines the number of clusters.
• To improve the accuracy of SAR image segmentation by integrating statistical modeling and spatial context.

Main Methods:

• Utilized a Gamma mixture model (GaMM) to define the marginal probability distribution of SAR images.
• Incorporated a Markov Random Field (MRF) to model spatial relationships between neighboring pixels.
• Implemented an iterative component merging strategy to reduce the number of clusters and estimate model parameters.
• Optimized segmentation by maximizing marginal probability and minimizing global energy.

Main Results:

• The proposed algorithm successfully determined the true number of clusters in both simulated and real multilook SAR images.
• Achieved more accurate segmentation results compared to existing algorithms.
• Demonstrated robustness in handling the statistical characteristics of SAR data.

Conclusions:

• The developed algorithm effectively segments multilook SAR images by adaptively determining the number of clusters.
• The integration of GaMM and MRF provides a powerful framework for SAR image analysis.
• This approach offers a significant advancement in the field of SAR image segmentation.