关键词: 模糊保险箱, 生物特征, 几何哈希法, 自动校准, 指纹

Abstract: In the concrete implementation of fuzzy vault algorithm, geometric Hashing is a kind of common technology for the biometric template automatic alignment. To solve the fuzzy problem of parameters selection in the algorithm implementation, this paper studies three parameters which affect the matching accuracy: image pixels size, the number of basic points and quantitative parameters of a Hash table (α and β). The optimal ranges of three parameters are obtained by carrying out the factor experiment analysis. Then, the extracting range of the minutiae algorithm and the rule of selecting the distance of basis points can be further optimized. Finally, the matching accuracy before and after optimization is compared and validated by the fingerprint picture based on the FVC databases. The experimental results show that the proposed optimization scheme can improve the matching accuracy of the algorithm because the FRR (false rejection rate) is reduced by 9.84%, at least FAR (false acceptance rate) is reduced by 7.12%, and has a certain robustness and practicality.

Key words: fuzzy vault, biometric, geometric hashing algorithm, automatic alignment, fingerprint