Characterization of ENVISAT multipolarization SAR data with bidimensional statistics

This paper is about evaluating the interest between studying directs dual images statistical indices and the use of statistical indices build from two SAR multipolarization images. We show an original way of using this kind of images by using two dimensionned statistical calculation in a Markov random field classification. We present results of our segmentation on urban, forest and mangrove areas. Those results are computed from ENVISAT ASAR data in HH and VV polarizations in a tropical context, near Libreville.