Obtaining Feature Correspondences

A state-of-the-art system for finding objects in images has recently been developed by David Lowe. The algorithm is termed the Scale-Invariant Feature Transform (SIFT) and intends to detect similar feature points in each of the available images and then describe these points with a feature vector which is independent of image scale and orientation. Thus feature points which correspond to different views of the same object should have similar feature vectors. If this process is successful then we should be able to use a simple algorithm to compare the collected set of feature vectors from one image to another in order to find corresponding feature points in each image. Figure 1 provides a typical example of using SIFT to locate a query image within a search image. The SIFT algorithm may be decomposed into four stages: