An Unsupervised Wishart Classification for Fully Polarimetric SAR Image Based on Cloude-Pottier Decomposition and Polarimetric Whitening Filter

Cloude-Pottier decomposition has been widely used in feature extraction and classification because it can include all scattering mechanism and can ensure the same eigenvalue under different polarimetric bases.A new unsupervised classification method is proposed for fully polarimetric SAR image based on Cloude-Pottier decomposition and polarimetric whitening filter(PWF).The result of PWF is used to improve the classic Wishart H/α classification instead of anisotropy A.As initialization,the Wishart H/α classification result is divided into 16 classes according to the PWF result.Then after another Wishart iteration,the final result would be got.The experimental results show that the PWF result has more information that the H,α and A parameters don't contain.It is also shown that the proposed classification method provides better performance than the general Wishart H/α classification and Wishart H/α/A classification.