Affine-invariant SIFT descriptor with global context

The paper proposed a new image matching method combining affine-invariant SIFT feature descriptor with global context. Affine-SIFT (ASIFT) can find large numbers of correct matches even with great variation of camera axis orientation, however when there are several similar regions or objects between the query and search images, it may also produce many mismatches. Therefore, we intend to introduce global context to resolve ambiguities that can occur locally when an image has multiple similar regions. A global context vector that adds curvilinear shape information from a much larger neighborhood, thus reducing mismatches when multiple local descriptors are similar. We have tested our method on various images and compare matching accuracy between the ASIFT descriptor with global context to that without.