Application Research On Distributed Arithmetic Coding In Biometric Authentication

Biometric authentication has been widely used in the area of information security as it has the advantages of safe,reliable,convenient and efficient.However,there always exists the risk that the original biometrics of users may be stolen when the typical biometric authentication system is attacked.Additionally,since the traditional DAC decoder based on breadth-first search has the problems of higher probability of misdeletion for the correct decoding paths and slower decoding speed,which causes the typical biometric authentication system adopting DAC has higher false reject error and longer authentication time.In order to solve these problems,this thesis focuses on the biometric authentication system based on DAC and proposes a new DAC decoder on the basis of depth-first search which is applied to the proposed biometric authentication system.The main works are as follows.(1)A biometric authentication model based on DAC is designed.As the first time a legal user employs the system,his biometrics will be taken out by the feature extraction module and transferred to the template library registration module.This module uses a DAC encoder to compress and encode his biometrics,and a check code along with the processed biometrics will be registered into the template library.Whenever a user tries to enter the system,his biometrics will be read by the feature extraction module and transferred to the legality verification module.This module adopts a DAC decoder to decode the encrypted template stored in the template library according to the biometrics of the user whose identity will be verified by the check code.(2)A decoding algorithm of DAC based on depth-first search is proposed.Given that the traditional decoding algorithm uses partial path metric to measure the decoding branches,the traditional decoder has slower decoding speed and higher code error rate.At the same time,the depth-first search method may result in different length of decoding paths,so the proposed DAC decoding algorithm comprehensively considers the effect of the extrinsic,intrinsic feature of the DAC bitstream and the length of decoding paths on path metric.A new computing method of overall path metric is designed leading to partial instead of global search of the decoding tree and consequently the search volume is reduced,which results in faster decoding.(3)Comparisons are made between the traditional decoding algorithm and the proposed one on decoding performances and authentication results by simulation experiments.The decoding performances are compared separately from decoding accuracy and speed and the authentication results are compared by the distribution of the receiver operating characteristic curves of the decoders.The experimental results show that under the same constraints,the two algorithms have no significant differences in decoding accuracy,but the decoding speed of the proposed algorithm is faster than the traditional one and the false reject error of it is 4.56%~5.65% lower than the traditional one.