An invariant approach for automated image registration of multitemporal remotely sensed imagery

Abstract

In this paper, a new automatic control point selection and matching technique for satellite image registration is proposed. The characteristic of this approach is that it uses features based on image moments and invariant to symmetric blur, scaling, translation, and rotation to establish correspondences between matched regions from two multitemporal images. The automatic extraction of control points is based on an edge detection approach and on local similarity detection by means of template matching according to a combined invariants-based similarity measure. The final transformation of the sensed image according to the selected control points is performed by using the thin-plate spline (TPS) interpolation. The proposed technique has been successfully applied to register multitemporal SPOT images from urban and agricultural areas. The experimental results demonstrate the efficiency and accuracy of the algorithm which have outperformed manual registration in terms of root mean square error at the control points.