EXPLORE PUBLICATIONS BY COUNTRIES

Biometric Template Security, Template Invariability, Template Quality Analysis

A biometric security is a technique by means of which digital contents are protected by a cryptographic key generated from the biometric features of a person like Retina, Iris, Fingerprint, Face, Voice and so on. Normally the digital contents like documents are protected by a cryptographic key generated from a unique password. The process in irreversible, i.e the key can be generated from the password but not the vice versa. Passwords are relatively easy to hack as most of the users keep their personal information like date of birth as password and also password length has a limit as human beings cannot remember a password of significantly large length. Hence guessing the password of a user, whose significant information is available, is easier. Therefore off late lot of emphasis has been given to biometric features. Biometric features of no two people are same. For example the finger prints or the face of any two people differ. Hence if a template (alphanumeric or binary representation of features from a biometric data) is selected for the key generation than cracking them for accessing information becomes significantly difficult. But as with every advantage comes certain limitations also. The keys are not time invariant. Templates tends to change based on the data acquisition, or with time. For example the finger prints or palm prints changes with ages. Iris, retina and face features changes with change in light intensity during the acquisition phase. Fingerprint features changes with change in the orientation of the finger while scanning. In a classic authentication problem, such variability’s can be easily dealt with by keeping a threshold for the acceptance of the features. Such acceptance threshold is not applicable for the case of biometric templates. Even slightest of the variability in the templates changes the generated key, therefore causing a high false rejection rate. Hence in this work we analyze the most accepted biometric features and techniques for key generation and propose the most invariable technique in terms of data acquisition invariability. The work analyzes Iris, Face, Fingerprint and Palm prints for analysis of the biometric template generation and key generation form the templates. Further a unique benchmark analysis technique is proposed for quantifying the quality of a biometric model or features.