Super resolution based on scale invariant feature transform

In this paper, SIFT (scale invariant feature transform) algorithm is used for the image registration of super resolution to ensure a more stable and accurate registration result, and thus improve the result of super-resolution which will be realized by least squares minimization. The advantage of this approach is that the super-resolution process will have a stable result even under severe transformation conditions. SIFT method is compared with Kerenpsilas method to prove its accuracy and robustness. Fine reconstruction results are also given to show the effectiveness of this approach. Simultaneous registration method can be introduced in future work to further improve the registration accuracy.