Rotation of polarimetric matrices and its effects on classification accuracy of man-made structures by synthetic aperture radar

This paper presents the effects of rotation of polarimetric matrices on man-made structure classification by polarimetric synthetic aperture radar (POLSAR). Both coherence and covariance matrices are used for classification. Some important modification is also applied to the previously suggested rotation algorithm. The advantage of this modification is to guarantee that the cross-polarized component is surely minimized. As a result, we were able to distinguish man-made structures from other areas more effectively than the related methods.