The newly emerging hand vein recognition technology has attracted remarkable attention for its uniqueness, noninvasion,
friendliness and high reliability. It is unavoidable to produce small location deviation of human hand in the
practical application; however, the existing recognition methods are sensitive to the hand shift or rotation. The test
sample is matched with a series of registered images after affine transformation including the shift or rotation by most of
researches, this affine transform method can remedy the location deviation to some extent, but the limited range for hand
shift and rotation brings users much inconvenience and the computational cost also increases greatly. Aiming at this
issue, a hand vein recognition algorithm based on local SIFT (Scale Invariant Feature Transform) analysis is developed
in this contribution, which has practical significance due to its translation and rotation invariance. First, the hand vein
image is preprocessed to remove the background and reduce image noises, and then SIFT features are extracted to
describe the gradient information of hand vein. Many one-to-more matching pairs are produced by the common
matching method of SIFT features, thus the matching rule is improved by appending a constrained condition to ensure
the one-to-one matching, which is achieved by selecting feature point with the nearest distance as the optimal match.
Finally the match ratio of features between the registered and test images is calculated as the similarity measurement to
verify the personal identification. The experiment results show that FRR (False Rejection Rate) is only 0.93% when FAR
(False Acceptance Rate) is 0.002%, and EER (Equal Error Rate) is low to 0.12%, which demonstrate the proposed
approach is valid and effective for hand vein authentication.