The most popular model-based decompositions (MBD) are reconsidered in the context of the estimation theory. It is shown that a large processing window is required to reduce the bias on the individual scattering contribution due the target scattering Reflection symmetry assumption. This limits the MBD efficiency in areas of non stationarity radar backscattering, such as urban areas. Eigenvector-based decompositions (EVBD) are also reconsidered. The effect of polarization basis change on target scattering description is investigated using the Cloude-Pottier ICTD, and the Touzi decomposition. The latter, which is an extension of the Kennaugh-Huynen method to partially coherent scattering, was originally developed to correct for the Cloude-Pottier scattering decomposition ambiguities. It is shown that the Touzi decomposition is the EVBD that permits the decomposition of both coherent and partially coherent scattering in terms of unique and polarization basis-invariant parameters.
展开▼
