The Research Of Biometric Encryption Method Based On Fingerprint

The biometrics are widely used in the identity authentication system because of their unique and stability.However,biometric templates cannot be revoked and republished as easily as classic keys.In order to ensure the privacy of users,the biometric cryptosystems technology formed by the combination of cryptography and biometric authentication came into being.At present,biometric cryptosystems based on helper data are the most widespread in application,and the fuzzy vault and fuzzy commitment are the maturest practical solutions.Nevertheless,the existing cryptosystems still have some loopholes,for instance,the traditional fuzzy commitment cannot resist cross-matching attacks and so on.In this paper,taking the fingerprint encryption as an example,we improve the traditional fuzzy vault and fuzzy commitment scheme.The main work of this paper is as follows:1.This paper expatiates on the classification and basic model of biometric cryptosystems,and introduces the traditional fuzzy vault and fuzzy commitment scheme.At the same time,the scheme of fingerprint feature extraction is implemented based on fingerprint encryption as a focal point,and the center location is designed according to the fingerprint classification.Then,transform the coordinates by using the extracted center as the coordinate origin to achieve the fingerprint alignment.We perform the simulation experiment with the images in database DB1_A of FVC2004 and show the fingerprint effect map after the image processing algorithm.2.To achieve the rapid fuzzy vault program.Firstly,we set the feature points credibility according to the gray image variance and the orientation map around the feature points and solve the problem where the fuzzy vault is time-consuming in the decryption stage.Then we build the transform function based on the user's password and realize the first layer encryption,which avoids cross-matching attacks.In this way,we form the revocable fingerprint template and achieve a both fast and secure fuzzy vault program.The simulation results show that the GAR and FAR are 94.2% and 0.11% respectively when the order of polynomials is 8,and the GAR performance and FAR performance are optimized by 3.2% and 0.1% compared with the original fingerprint fuzzy vault algorithm.3.We design the method of coding the sector feature of fingerprints.The fingerprint image is segmented into different blocks by rays and concentric circles.Combining the number,type and direction of the minutiae in each block,we complete the correspondence between the fingerprint feature and the binary ordered sequence.We also improve the fingerprint's fuzzy commitment program.First of all,utilize the error correction code to yield the spatial transformation for fingerprint feature sequence and achieve a secure mapping process.Secondly,with the error-correction performance of BCH code,the requirement of fuzziness in biometric authentication is completed.By selecting different error-correction capabilities,the security level of the identification system can be dynamically adjusted.In this paper,we use the short coding and long coding scheme to generate the fingerprint code,and BCH(15,7,2)and BCH(15,5,3)are used for error correction at the time of decoding.When the matching score is set to 70,GAR and FAR in the short coding scheme can achieve 93.8% and 0.75% respectively,and GAR and FAR in the long coding scheme achieve 91.6% and 0.57% respectively.