Title: A New Paradigm for Polarimetric SAR Data Analysis
Abstract: Under the Stokes formalism, a 4 x 4 real matrix, i.e., Kennaugh matrix represents a PolSAR observation. Under such a scenario the geodesic distance (GD) on the unit sphere in the space of 4 x 4 real matrices forms a useful measure to compare normalized Kennaugh matrices. This has many excellent properties like boundedness, invariance under scaling and orthonormal transformation of the wave polarization basis. Recently Ratha and Bhattacharya have exploited this convenient measure for a number of applications such as: change detection, unsupervised classification (clustering), establishing indices such as Radar Built-Up Index (RBUI) and Generalized Radar Vegetation Index (GRVI) in the context of the study of built-up area extraction and vegetation monitoring from PolSAR images respectively. Additionally, a generic Scattering Power Factorization Framework (SPFF) has also been proposed as an alternative approach to modern model-based PolSAR target decompositions. These techniques are utilized with full-polarimetric RADARSAT-2 and ALOS-2 datasets.