Unsupervised Classification of Polarimetric SAR Images Based on ICA
Haijiang Wang, Yiming Pi, Zongjie Cao · 2007
Polarimetric SAR image classification is an important research area. Various classification methods continue to be developed for specific applications. In this paper, A new unsupervised classification method for polarimetric SAR images is proposed. It is based on independent component analysis (ICA). By ICA processing, several independent components are extracted from the channels of the SAR images. One of the independents is regarded as speckle noise and thrown away. By taking each remained independent as a kind of target, a classified SAR image with higher classification accuracy can be obtained.