Biometric Template Public Protection And Template Parameter Management Research

In biometric recognition system, the template protection involved in the template and the template parameter is the key to the safe of recognition system. However, the template generation algorithms are insufficient in meetting the four characteristics and resist attack problems when they published templates, and the parameter management schemes have single scene in biological system.This paper has a research on the template protection, and proposed a parameter management scheme based on secret sharing in the point of parameter management, and combined with the Biohashing algorithm to design a scheme for the purpose of identification. The main researchs of this paper are as follows:1. This paper have a study on the open template protection. Firstly, this paper point out the existing problems of Biohashing algorithm and fuzzy vault at the template protection to meet the four characteristics. From the view of parameter management, this paper introdue the four parameter management, and analysis their weakness on the template protection for the further study.2. In the point of the storage of template, this paper designed a parameter management scheme based on the secret sharing algorithm. Firstly, this paper constructed a polynomial to finish the distribution of parameter. Secondly, this paper used the Lagrange iAterpolation method to reconstruction the parameter. Thirdly, this paper designed an experiment to have a research on the length of parameter and (t, n)'s influence to time overhead in the distribution and reconstruction of parameter. This paper selected different length of the parameter and (t,n) values to have some simulation experiments on the platform of Matlab, the results and analysis shows that the parameter is safe and the time cost is good to meet the needs of feasibility.3. In order to reduce the security of Biohashing in the identification scheme, this paper combined with the parameter management and Biohashing to design an identification scheme. Firstly, this paper published and reconstructed the random number seeds original stored in token. Secondly, this paper designed a research on the effect of the number m to the time overhead. Thirdly, the average time of seed distribution and reconstruction was calculated by 1000 experiments with different m values on the platform of Matlab. The results and analysis shows that the optimal range of the number m of the seed vector is 20?200, which effectively reduces the fake attacks of the original Biohashing algorithm, and implements a variety of landing mode of the user.Finally, according to the simulation results, this paper seclet the optimal value of (t, n) and m to study the recognition rate and speed of scheme in the MATLAB platform, the results show that the size of FAR, FRR and average. The analysis of data shows that the identification scheme solves the single usage scenarios, improves the safety parameters and realize the template of the indirect, to meet the users are more simple, more convenient, more intelligent life philosophy.