An Improved Four-Component Model-Based Decomposition Scheme with Emphasis on Unitary Matrix Rotations

Abstract

Scattering mechanism ambiguity has been a significant challenge in the field of model-based decomposition of polarimetric SAR data. Even after continuous reported advancements, still, it is not being concluded that problem have successfully been suppressed. To address this issue, the proposed method focuses on the analysis of specific complex urban and sloped mountainous bare land profiles that can rotate the polarization basis. The approach optimizes the coherency matrix by subtracting helix component prior to decomposition followed by the incorporation of unitary matrix rotations to decouple the energy between the orthogonal states of polarization by neutralizing T23 and T13, separately. Furthermore, instead of conventional branching condition, mean alpha angle had been utilized to discriminate between dominant surface and dihedral scattering area. Validation has been done using two different polarimetric datasets. Quantitative analysis shows the improved decomposition results through empowering the co-polarized powers in their respective underlying dominant scattering areas.