Fig. 7 shows binary change detection maps derived from polarimetric
similarity tests such as the likelihood-ratio test statistic in
the Wishart distribution, Q, and the covariance matrix similarity,
R. In this case, only two classes (unchanged ωn and changed ωc) are
considered based on Eqs. (5a), (5b), (5c)–(7). In farmlands, the result
derived from the Wishart test is similar to the results based on ΔHH
and ΔVV, however the Wishart test exhibits better detectability in
damaged urban areas. On the other hand, the similarity measure
provides the best results in detecting collapsed buildings without
false-alarms in undamaged areas, such as seasonally changed paddy
fields. For this reason, the similarity parameter identifies scattering
mechanism differences between two covariance matrix acquisitions
regardless of changes in the total power, which can be sensitive to
seasonal changes.