Hand Vein Recognition Based On Multi-feature Fusion

Since the1960s, the information technology developed rapidly. The importance ofinformation security has reached an unprecedented peak. The traditional identificationmethods (such as keys, passwords, etc.) has been more easily fabricated. In this case,biometric has been further studied. Biometrics is a science. The characteristics (e.g fingerprint,iris, DNA, vein) and traits (e.g voice,gait) are the researching target of biometric. Due to theadvantage such as living body identification, noncontact, unable to forge, hand veinrecognition has been warmly welcomed by all researchers.In this paper, the hand vein recognition is seen as a specific research object for biometricrecognition technology. We have made basic research and scientific summary to theapproaches of hand vein recognition technology. The every component: capture device,preprocessing, feature extraction and feature fusion is discussed. The focus is a new hand veinrecognition method based on feature fusion.The main focus of this paper are as follows:1. A hand vein image photograph equipment is set up. This equipment has simple structure,and will cost less time to capture, the authentication process is simple, fast, unable to fabricate,noncontact and high value.2.A new approach of hand vein recognition is presented. The calculation will be reduceddue to the concept of reference points. The recognition algorithm based on the characteristicsof the respective connection angle between reference point and feature points can overcomethe effects cause by image translation and rotation. The results indicate that the algorithmmentioned above do achieve hand vein recognition reliably and quickly.3.Multi-feature fusion system at decision layer is discussed. Moment Invariant representthe global features of the image. The angles between corresponding connections are detailfeatures. Both global and local features will be made full use by the fusion of the two featuresmentioned above. Compared to single feature recognition, there will be less false reject rateand false accept rate using multi-feature fusion recognition. There will be an actualapplicative value and a promising future.