Selective corner matching for high-resolution image registration
Sultan Alkaabi, Farzin Deravi · 2006
An efficient corner matching scheme in high-resolution images is presented. The approach partitions the two images into equal sub-images, and each sub-image of the first image is linked (paired) with all highly similar sub-images from the second image. The unlinked sub-images are suppressed from further processing. This limits the number of hypothesis-matched corner pairs that are processed, as only linked sub-images that sufficiently overlap are searched for correspondences. Experiments are presented to show the effectiveness of the proposed scheme.