A SAR-SIFT Like Algorithm for PolSAR Image Registration

The synthetic aperture radar scale-invariant feature transform (SAR-SIFT) algorithm has been applied to synthetic aperture radar (SAR) image registration and has good performance. However, compared with single polarization radar images, polarimetric synthetic aperture radar (PolSAR) images have more polarization combinations, which can obtain more polarization information. Therefore, based on the scattering mechanism of PolSAR, this paper proposes a registration algorithm suitable for PolSAR images. The algorithm improves the SAR-SIFT algorithm and uses the polarization information of the PolSAR images to introduce a new gradient definition and construct new scale spaces. The algorithm proposed in this paper is tested on multi-time and multi-angle PolSAR images. The experimental results show that the proposed method can make feature points more widely distributed and obtain more correct matching point pairs, which effectively improves the registration performance of PolSAR images.