Abstract:

An unsupervised classification method using Freeman decomposition and fuzzy clustering is proposed to solve the ambiguity problem among surface, volume and double bounce scattering dominated region, which is extracted from  polarimetric synthetic aperture radar (PolSAR) data with Freeman decomposition. A fuzzy clustering method of polarimetric interferometric SAR (PolInSAR) data making use of two parameters  describing  optimum coherence which are optimum coherence entropy  H Int and optimum coherence anisotropy  A Int is proposed to partition different scattering mechanisms dominated region. Fuzzy theory is introduced to the partition of  H Int / A Int plane to get intial partition of the image. Then cluster merging and fuzzy clustering operations are introduced to obtain the final classification result. Experiment results making use of full polarimetric interferometric data of Oberpfaffenhofen area acquired by ESAR confirm the validity of the presented method.