Research On The Encryption And Recognition Algorithm Of Finger Vein

Biometric Authentication Technology overcomes the shortcomings of traditional authentication technology and has been widely used in this information age.As one of the most important technologies of authentication,Finger vein recognition attracts our attention because of its high security,reliable accuracy and excellent performance.However,there are two important disadvantages in the current finger vein recognition system.One is that the user's feature vectors are saved in the database directly without any encryption which would cause a lot of security problems.The second is that the pre-process of finger vein image is too complicated and its robustness is very poor.So this paper studied these two problems and proposed some solutions.The main works of this paper are as follows:(1)In order to solve the security problem of finger vein template,this paper designed finger vein encryption scheme based on fuzzy fault.According to the finger vein minutiae,this scheme generated genuine set G and chaff point set C by polynomial mapping and mixed them together to form the fuzzy vault,which can protect the genuine set G effectively.(2)This paper also proposed the finger vein encryption scheme based on fuzzy extraction,we regard inherent biometric fuzziness as additive noises in communication systems and extract finger vein feature by MB-CSLBP algorithm.Then Reed-Solomon Code is used to narrow the differences within intra-class and only public information and random key are saved in the database.(3)In addition,we did experiments based on these two encryption schemes and analyzed their anti-counterfeiting performance,re-release performance and irreversibility.Also we compared advantages and disadvantages according to their performance.(4)In order to solve the problem of complicated image pre-process,a finger vein recognition method based on the convolution neural network is proposed in this paper,and the equal error rate(ERR)of this method reached 0.21%.Compared to the traditional finger vein feature extraction,this method not only reduces the number of pre-processing steps and improves the recognition accuracy,but also shows great robustness in both illumination and rotation.