An Adaptive Polarimetric SAR Speckle Filtering Algorithm Based on Unsupervised Classification

Abstract

In this paper, a new approach is proposed for speckle reduction in full polarimetric SAR images. Being different from others, it adopts an enhanced unsupervised classification method to estimate the parameters of the speckle filter and thus an adaptive moving window is formed to better speckle reduction and preservation of texture information. . Experimental results demonstrate the effectiveness of this new filtering algorithm. Comparisons with other methods are also made.