Angular similarity based cancelable biometrics authentication for remote/cloud applications

Abstract

Data privacy is highly required for remote or any cloud storage applications. Several security measures are available for protecting data but there are chances for data breaches. To overcome the causes, biometric systems are widely used for enhancing security. The biometric databases have to be protected to avoid malicious attacks. Cancelable biometric system is one such method which transforms the features of biometric and make repository so that the actual trait is protected. Cancelable feature vector are numeric which helps for both identification and verification task. The challenge is to provide authentication using cancelable biometric. The cancelable biometric have to be recreated or any authentication metrics such as histogram analysis, ROC etc. have to be used. The proposed system uses angular similarity metrics for authentication. Here, similarity is measured between actual feature vector and cancelable vector to set up the threshold. Based on the threshold value, authentication is performed, which proves to have 98.2% accuracy with less computation time and reduced FAR.