Algorithm of Image Stitching Based on SIFT Feature Matching
Tao Guo · Computer Technology and Development · 2009
A fast and accurate image stitching algorithm based on SIFT feature matching is proposed.SIFT feature matching algorithm is hot and difficult in the field of matching feature points research at home and abroad.Its match ability is strong,and can be dealt with translation,rotation,affine transformation circumstances matching problem between the two images.To some extent,images also have a more stable feature matching capacity even in the perspective of arbitrary capture.In this paper,a new image stitching algorithm is proposed on the basis of this match thinking and multi-resolution spline.It uses multi-resolution spline to determine the width and splicing regional weighting function,so that images are blended at different resolutions,thus the blended-image is clear and smooth seamless.Algorithm in most cases can be automatically completed,the experimental results show that it improves the stitching speed and reduces the difference of image brightness.It makes the blended-image natural and realistic.