Abstract:

There are obvious crack defects in most of the existing algorithms in the process of image mosaic, therefore we present a best point smoothing method for finding the best seam. Firstly, we extract feature points through the scale invariant feature transform (SIFT) algorithm and reduce the dimension of the feature descriptors with the PCASIFT algorithm. Then we eliminate the false matching points by the RANSAC and take the closest points to the three primary colors as the splice points. Finally the splice point set is smoothed and the best stitching line is obtained. Simulation results show that the proposed  algorithm has high time efficiency and can obtain a better color image mosaic.

Key words: feature points;PCA-SIFT descriptor;image registration;image stitching