Facet based differential registration of remote sensing images

Registration and fusing of images are important in the synthetic application of multi-source remote sensing images, and the former is the base of the latter. Good results can be acquired in a flat region by the traditional method. However, the result is not usually good in a mountainous region using the same method, and it leads to bad image fusing. A new registration method, called "facet based differential registration" of remote sensing images, is presented. The key strong point of the method is to acquire dense control points (CPs). A dense irregular triangle network (TIN) is generated from the dense CPs. The relative registration is based on the facet of the dense TIN. The fine registrated images are assurance of fine fusing image. Some experiments using many practical examples demonstrate that the new method is suitable for registration and fusing of remote sensing images from not only mountainous regions, but also flat regions