Remote sensing image registration based on local structural information and global constraint
Yue Wu, Wenping Ma, Qingxiu Su, Shaodi Liu, Yuhuan Ge · Journal of Applied Remote Sensing · 2019
In the process of remote sensing image registration, a robust point-matching algorithm is very important to get better results. However, it may be difficult to find enough correct correspondences when remote sensing images exhibit various noises and local deformations. Although existing methods can remove the outliers, many inliers will be removed at the same time. We propose a point-matching method for remote sensing image registration based on phase congruency normalized cross-correlation and global constraint. First, the classical synthetic aperture radar scale-invariant feature transform algorithm is used to obtain the initial correspondences. Then, the feature distance and phase consistency normalized cross-correlation are combined to remove the outliers in the local neighborhood blocks of the initial correspondences. Finally, the proposed method focuses on increasing the number of inliers and raising the correct ratio by a global constraint. The experimental results compared with several state-of-the-art algorithms show that our method can get more correspondences with higher accuracy.