Automatic feature based image registration using SIFT algorithm

Image registration is the process of mapping and geometrically aligning the two or more images. The steps in image registrations include: feature detection, feature matching and image transformation, and resampling. The accuracy of a registration process is highly dependent on the feature detection and matching. In this paper, we use a SIFT (Scale Invariant Feature Transform) algorithm to detect features, which is invariant to rotation, scaling and noise. Then initial matching is computed using Euclidean distance, and mismatch features between point pairs are eliminated using RANSAC. The result shows that the automatic registration algorithm is correct and effective.