Performance Analysis of Template Data Security and Protection in Biometric Systems

摘要

Biometric systems are one of the most popular and accurate approaches for human identification or verification in modern computing. A biometric system works by collecting important biological or behavioural traits of the prospective user. Extracted traits are stored as a feature template in the database of the biometric system. The reliability and accuracy of the biometric systems are improved by using multiple modalities, sensors or algorithms during recognition process. These systems are called as Multibiomtric systems which results into greater accuracy but with increased cost and complexity. In comparison to unibiometric system, a multi-modal system provides better recognition performance and protection from spoof attacks. Biometric systems are generally used in modern day computing to provide security during authentication in many applications but these systems themselves are vulnerable to various types of attacks. In this paper, an analysis of various security issues and template security schemes for Biometric Systems is presented. One of the challenges is to protect user's raw image data from imposters as well as to secure template data. It is found that template data security is a major design issue and needs to be addressed by designing low overhead security schemes. Template security schemes are dependent on the overall performance parameters of the system and must be computationally hard to avoid modality reconstruction.