Data classification based on PolInSAR coherence shapes

Abstract

The combination of SAR polarimetry and SAR interferometry introduced many new research and application fields. Mostly they go into the direction of physical parameter in- version and media characterization and classification. The major parameter which is computed from observable of polarimetric interferometric SAR (PolInSAR) is the complex coherence. In most fields only a few coherences are used for their purposes. But coherences are easily influenced by many parameters, like temporal or spatial baseline, or frequency. In this paper we present a new numerical approach for calculation of the shape and distribution of all complex coherences, which are possible with equal scattering mechanisms. Based on these shapes, new parameters are introduced for data decomposition. Finally, the PolInSAR data is classified based on coherence shapes. The per- formance of our approach is tested on PolInSAR data in L-band, acquired by airborne ESAR system. The result is compared with the classical polarimetric entropy-alpha-anisotropy approach.