A Study of An Improved SIFT Descriptor

摘要

SIFT (Scale-Invariant Feature Transform) descriptor is an images' local feature descriptor and is invariant to image scale, rotation and translation. Besides, it's relatively robust to change in illumination, affine transformation and change in 3D projection transformation. This paper presents an improved SIFT descriptor. Experiments show that it can describe local features of objects more efficiently, accurately and rapidly.