An improved SIFT feature matching algorithm
Yuning Hua, Jing Lin, Chao Lin · 2010
For the diversity of feature extraction and the complexity of similarity calculation in the feature-based image registration methods, an improved Scale Invariant Feature Transform (SIFT) feature matching algorithm is proposed. First of all, by using the classic SIFT algorithm, the feature points of the images are extracted. By using the gradients normalized method eigenvector descriptor is formed. Then the feature points are matched according to the Euclidean distance ratio. At last, by using the bilateral matching algorithm, the mismatch points are removed. The experiments show that this method is reliable and practicable.