FRACTIONAL BROWNIAN-MOTION MODELS FOR SYNTHETIC-APERTURE RADAR IMAGERY SCENE SEGMENTATION

This paper demonstrates the application of fractal random process mode ls and their related scaling parameters as features in the analysis an d segmentation of clutter in high-resolution, polarimetric synthetic a perture radar (SAR) imagery. Specifically, the fractal dimension of na tural clutter sources, such as grass and trees, is computed and used a s a texture feature for a Bayesian classifier. The SAR shadows are seg mented in a separate manner using the original backscatter power as a discriminant. The proposed segmentation process yields a three-class s egmentation map for the scenes considered in this study (with three cl utter types: shadows, trees, and grass). The difficulty of computing t exture metrics in high-speckle SAR imagery is also addressed. In parti cular, a two-step preprocessing approach consisting of polarimetric mi nimum speckle filtering followed by noncoherent spatial averaging is u sed. The relevance of the resulting segmentation maps to constant-fals e-alarm-rate (CFAR) radar target detection techniques is also discusse d.