A Multi-sensor Image Registration Algorithm Based on Structure Feature Edges
一种基于结构特征边缘的多传感器图像配准方法

Image registration is prerequisite to multi-sensor image processing, such as image fusion. Taking synthetic aperture radar (SAR) images and optical images containing man-made objects as examples, this paper proposes a multi-sensor image registration algorithm based on structure feature edges. The common features for the same man-made objects shown in the two kinds of images, i.e. structure feature edges, are extracted firstly, and virtual corners are constructed on the basis of edge matching. The coarse registration based on feature consensus and the fine registration based on virtual corners are used to realize the automatic coarse-to-fine registration between multi-sensor images. Experimental results show that the proposed algorithm has a high registration accuracy.