A Fast and Robust SAR-Optical Image Registration Algorithm

In recent years, the rapid advancement of remote sensing technology has led to significant progress in the field of heterogeneous remote sensing image registration. However, most existing heterogeneous image registration algorithms primarily rely on structural features, gradient information, or phase consistency of the images for registration. This approach poses significant challenges in terms of registration accuracy and efficiency when dealing with remote sensing images of nonuniform texture scenes, such as areas containing oceans, deserts, and forests. To address these issues, this paper proposes a fast and robust image registration algorithm tailored for non-uniform texture regions. The algorithm first employs an adaptive regionaware feature point extraction method to distinguish between effective and weak texture regions, thereby enhancing the registration robustness for scenes with significant regional differences. Secondly, it utilizes a single-scale orientated gradient fusion operator to construct feature descriptors, which not only allows for the rapid construction of feature descriptions but also boasts high accuracy. Finally, accurate image registration is achieved through fast template matching in the frequency domain based on non-maximum suppression. The application of this algorithm to the registration task of heterogeneous images with non-uniform texture regions demonstrates significant improvements in both registration accuracy and efficiency.