A New Classification Approach for PolSAR Images

PolSAR images provide more information than the conventional radar images and so greatly improves the ability to terrain discrimination. This work is focused on the problem of the classification of different terrain types of PolSAR images. The proposed method uses a combination of two different classifiers. This method is named as Meta classifier and compared with other competitive classifiers. One as a spatial partitioning method, assigning each leaf of tree a distinct region label. So they are called as clustering trees. The other one is used as the second classifier for generating the class labels. But the final result is generated by voting over the ensemble results of the two classifiers. The experimental results obtained so far, indicate the potential of this approach.