Kaan Ersahin (Univ. British Columbia): “Segmentation and Classification of Polarimetric SAR Data Using Spectral Graph Partitioning”

Kaan Ersahin (University of British Columbia, Canada):
“Segmentation and Classification of Polarimetric SAR Data Using Spectral Graph Partitioning,”
November 2009.  Advised by Prof. Ian G. Cumming and Prof. Rabab K. Ward

Polarimetric Synthetic Aperture Radar (POLSAR) data is an important source for many operational remote sensing applications. Segmentation and classification of image data are important tasks for POLSAR data analysis and interpretation, which often requires human interaction. Existing strategies for automated analysis have only utilized the polarimetric attributes of pixels, including target decompositions based on physical, mathematical or statistical models. In this thesis, spectral graph partitioning methodology is used to exploit both the polarimetric attributes of pixels, and the visual aspect of the image data through visual cues. A new unsupervised classification algorithm is developed, where segmentation based on the contour and spatial proximity cues is followed by classification using the similarity of coherency matrices. Two data sets over agricultural fields acquired in L-band from AIRSAR, and C-band from the Canadian CV-580, were used to obtain the experimental results, which suggest improvements over the Wishart classifier. This methodology is flexible in the definition of affinity functions and will likely allow further improvements through additional image features and data sources.

Click here to access the thesis or contact the author.