Statistical modeling of multipolarization and multifrequency SAR images of the sea surface

A statistical analysis of polarimetric and multifrequency SAR images of the sea surface is presented. The analysis is carried out resorting to binary hypotheses tests aimed at verifying to what extent the texture value is shared between pixels of the same image in spatial proximity to each other, measurements of the same pixel acquired at different polarizations but in the same band, and measurements of the same pixel at different frequencies but with the same polarization. The data under test were acquired during a SIR-C/X-SAR mission in 1994. It is shown that cells in spatial proximity share the same value of the texture over sub-images of certain dimensions; moreover, while the HH and VV measurements of the same pixel possess a common value of the texture both in the L- and the C-band, the texture generally varies from the L- to the C-band (for both the HH and the VV channel).