A robust and adaptable high-precision method for matching flipped SAR images based on an oriented descriptor

Abstract

ABSTRACT The increasing imaging capability of spaceborne Synthetic Aperture Radar (SAR) systems makes a challenging task in matching slant-range (SR) SAR images acquired from different orbit and side-viewing directions. However, currently existing algorithms for the SR SAR images matching are unsatisfactory and the correct homologous points cannot be matched. In this work, a robust and adaptable high-precision method for matching flipped SAR images based on oriented descriptors is proposed. Firstly, the ratio gradient calculation based on Gabor odd filter is introduced. Secondly, the oriented descriptors are built to eliminate the diversity of feature-based descriptors of homologous points of flipped SR SAR images, enabling directly match the flipped SAR images. Finally, the feasibility and reliability of the proposed method are verified by the experimental results of simulated flipped SAR images and actual flipped SAR images, and two directional descriptors of the slave image are constructed to match non-flipped SAR images. The results show that the proposed method performs better in precision with the root mean square error (RMSE) less than 1 pixel, and the proposed oriented descriptors can match SR SAR images.