PolSAR image speckle reduction based on pixels classification

An effective algorithm for PolSAR image speckle reduction is developed in this paper. First, pixels in the image are classified through three steps: i) HH/α plane is used for image initial classification based on the polarization difference; ii) anisotropy A and total backscattering power after PWF are combined for further classification; iii) hierarchical cluster method is applied to get the final classifications. Then, the pixels are selected to be filtered according to their classes by the linear minimize mean square filter. Experimental results with AIRSAR data show that, the new algorithm is more effective than Scattering-Model-Based speckle filter, not only in speckle suppression, but also in polarimetric properties, structure and texture preservation.