In order to solve the problem of limited adaptability and arbitrary rigidity of H/a initial partition in classical unsupervised classification methods for polarimetric SAR images

a novel unsupervised classification method is proposed. Combined with Pauli Decomposition and Wishart Distance based on tatal polarized power.Firstly

the algorithm uses tatal polarized power Span

initial classification are carried out based on the scattering strength

and then combined with the initial the classification results and 3 bands of HH

HV

VV which obtained from the Pauli decomposition for iterative classification

finally classification results are obtained by clustering based on the Wishart distance. Validity of Pauli decomposition-based iterative improvement classification method are demonstrated in experiments using two groups of L-band fully Polarimetric SAR data of NASA-JPL Laboratory.Classification result are compared with traditional H/α-Wishart classification method

precision and rationality are improved.