Stitching Algorithm of Sequence Image Based on Modified KLT Tracker

Against to the low precision and efficiency of most current image stitching techniques, this paper proposes an algorithm for automatic stitching of sequence images. Extracting the feature corners from the image and selecting the most suitable ones to track, then these selected corners are regarded as moving objects, tracked and located accurately in all the sequence images with our modified KLT tracker. after that, all the overlapping relationship are confirmed, so we can select the most suitable mosaic frames and sort them by our filter rule. Finally, this paper proposes a fusion method that can fulfill seamless mosaic. Experiment results show that the precision and efficiency are both improved and the algorithm is robust.