The distribution of interferometric phase differentials and a self-initialising PolInSAR classifier

This paper describes an unsupervised classifier for polarimetric interferometric SAR (PolInSAR) data. Expectation maximisation is used to estimate class parameters that maximise the probability of a dataset for a given number of classes. Polarimetric information, in the form of coherency matrices, and interferometric information, in the form of complex coherences, is taken into account. Phase differentials between complex coherences in different polarisation bases are used to make the classifier sensitive to the vertical structure of the scene under observation, and a distribution of such phase differentials is developed. The classifier is self initialising in that it does not rely on decompositions or thresholds. Classification results based on real data are presented and discussed