Mode-Based Method for Matching of Pre- and Postevent Remotely Sensed Images

With the continuing increase in the number of images collected every day from different sensors, automated registration of multisensor/multispectral images has become a very important issue. This is particularly true when pre- and postevent image comparison is concerned: For this particular application, the requirement of obtaining the earliest possible postevent image imposes the use of data potentially showing strongly different characteristics with respect to the pre-event image. Strongly inhomogeneous image pairs require robust automatic registration techniques. Resolution-independent feature-based registration is naturally preferred over correlation-based registration where data are inhomogeneous. In this letter, we propose a mode-based feature matching scheme, formerly invented for computer vision application and adapted in this letter to pre- and postevent matching. We list a few weak points in the original technique when used for this particular application and illustrate how a significant improvement was obtained by modifying the algorithm. Three real cases of pre- and postevent feature matching on high resolution satellite images are shown and discussed.