An Improved Hybrid Freeman/Eigenvalue Decomposition For Polarimetric SAR Data

In this paper, the negative power problem in hybrid Freeman/Eigenvalue decomposition is analyzed. To solve this problem, an improved hybrid Freeman/Eigenvalue polarization decomposition method is proposed. Firstly, orientation angle is compensated and phase angle is rotated of coherent matrix through judging dominant scattering mechanism. Secondly, considering the continuous variation of HH and VV ratio, a generalized volume scattering model is introduced into the hybrid decomposition method. Finally, negative power problem in the hybrid Freeman/Eigenvalue decomposition method is further solved by using the nonnegative eigenvalue method. The results are verified by using ESAR data from Oberpfaffenhofen area in Germany. The results show that this method can effectively solve the problem of negative power and the decomposition result is easy to be combined with the actual feature.