Elastic registration of remote-sensing images based on the nonsubsampled contourlet transform

Image registration plays a critically important role in remote sensing applications. Due to the large volumes of remote-sensing data available today, automated registration of multitemporal and/or multisensor images is highly desired. In this work, a new automatic approach for elastic image registration of remotely sensed images is proposed. The critical elements for an automated image registration procedure are explored. The elements include control-point (CP) extraction, CP matching, and transformation parameters estimation. In the proposed algorithm, a new CPs extractor has been developed. This technique exploits the nonsubsampled contourlet transform (NSCT) to automatically extract a set of CPs where misalignment between images can be expected to appear. The proposed algorithm has been successfully applied to register multitemporal ALSAT-1 images from urban and agricultural areas. The experimental results demonstrate the robustness, efficiency and accuracy of the proposed algorithm.