descriptor matching

Pairing each keypoint descriptor in one image with its closest-distance match in another.

Descriptor matching pairs each keypoint descriptor in one image with the descriptor closest to it (by Euclidean distance for SIFT-style descriptors, Hamming distance for binary descriptors) in a second image's descriptor set, producing a set of candidate point correspondences a downstream geometric fit (homography, fundamental matrix) can then use.

Examples

Every SIFT descriptor in image A is paired with whichever descriptor in image B sits closest to it by Euclidean distance, producing a first pass of candidate correspondences before any filtering.