The new curvature methods of image fine registration for synthetic aperture radar interferometry

The image fine co-registration whose accuracy has a direct influence on DEM accuracy, is more important for synthetic aperture radar interferometry (INSAR). The co-registration based on intensity values is easy to implement and computationally efficient but the least accurate. To get better accuracy, we must consider the characteristics of the complex images intensity, the geometrical characteristics of the distributed targets, and the correlative characteristics between the two characteristics. This paper presents the curvature method for fine co-registration of the images. It makes use of not only the intensity value of the pixel, but also the geometric distributing of the group of pixels. At the same time, the characteristic surface is built by combining the two characteristics. The curvatures of some curves in surface show the geometric characteristics of the surface, which have no relation with the surface flexibility and its coordinate position. The characteristic of the registration objects is obvious when using the curvature of surface as the fine co-registration object. This method is feasible and easy to be implemented. Detailed theoretical deduction and practical research is done in this paper. The experiment result shows that the registration accuracy is better than that of traditional methods.