A New Remote Sensing Image Registration Approach Based on Retrofitted SIFT Algorithm and a Novel Similarity Measure

Through optimizing the feature matching process by integrating scale-invariant feature transform (SIFT) algorithm with a novel curve matching algorithm based on isohypse and triangle-area representation (TAR), two gaps of SIFT when applied in multimodality image registration are supplied. In this paper, a improved similarity measure (SMLF) based on trajectories generated from Lissajous figures is also proposed. Based on this similarity measure and integrating with noticeably improvement on feature matching process by using the edge information, a new registration approach is constructed, which is more robust and accurate than prior approaches.