SIFT algorithm based on the large-format image block

By studying the Scale-Invariant Feature Transform (SIFT) operator, we know the process require calculate the whole scale spaces of the input image, costing the expensive time and taking up a lot of memory resources. Especially for the large aerial image, the SIFT operator can not process and doesn´t generate keypoints. Therefore we implement the SIFT operator on large-format aerial images, and show that it can quickly and accurately detect keypoints in the large-format aerial image for automatic matching.