Desain Metode Polar Fuzzy Vault untuk Proteksi Data Sidik Jari

Abstract

Authentication concept using password or token has been used widely but still has weaknesses, because password or token can be easily forgotten or shared with other users. The use of biometric data as an authentication tool can be a solution to the problem. It is because biometric data is relatively permanent and unique, thus, the risk of losing or forgetting the password is reduced. Consequently, a protection method to secure the biometric data is needed. A number of studies have been done to secure the biometric data, ranging from classic encryption method to data transformation.
In this paper, a method is developed to create a cancelable fingerprint templates by combining pair-polar on the selection and matching with modified fuzzy vault on data transformation. The selection procecss is carried out in two stages. The first stage selects the reference minutiae points and its respective neighbors which close to the reference. A feature vector set is adopted from pair-polar method with an addition of the ridge type of the neighbors for later use in the matching process. The second stage generates a template, which is developed by implementing extended fuzzy vault with combining four regular fuzzy vault sets into one vault. This method gains the most optimum accuracy up to 99% of GAR and 0.03% of FAR. The use of ridge type in the matching process can reduce FAR from 1.03% to 0.03%, and the use of extended fuzzy vault can reduce FAR from 0.28% to 0.03%.