Computer Vision: SIFT algorithm (Scale Invariant Feature Transform)

The Scale Invariant Feature Transform (SIFT) is a computer vision algorithm used to detect and describe local features in images for object matching. It achieves scale invariance by constructing image pyramids, or octaves, with varying levels of Gaussian blur and then identifying keypoint extrema within the Difference of Gaussians (DoG) space. To ensure rotation invariance, SIFT assigns an orientation to each keypoint based on local image gradient directions. A 128-dimensional feature descriptor is then computed for the region around each keypoint. These descriptors can be compared across different images using metrics like L2-distance to find corresponding points and match objects despite changes in scale, rotation, and illumination.