Scattering power decomoosition using fully polarimetric information

There exist 9 independent polarization parameters in the coherency or covariance matrix as the second order statistics. Various decomposition methods have been presented based on the physical scattering model using these parameters. However, none of them accounts for all polarimetric information, typically leaving T₁₃ element un-accounted in the coherency matrix. This paper presents a complete four-component scattering power decomposition method using all polarimetric information. Using double unitary transformation of measured coherency matrix, it is possible to eliminate T₂₃ element, which results in a reduction of polarization parameters from 9 to 7. Then by unitary transformation of expansion matrices, it becomes possible to account for T₁₃ term, which has never been accounted for in the physical model-based decomposition. By the double unitary transformations to minimize the T₃₃ component, all polarimetric parameters are accounted. This methodology also reduces the negative power problem significantly. This method is a further extension of the existing four-component decomposition. The four scattering powers (surface, double bounce, volume, helix) are assigned to blue, red, green, and yellow to compose full-color image. An example image of San Francisco area acquired with ALOS-PALSAR is shown to validate the decomposed result.