Frame work for fake face identification using hybrid approaches

Abstract

Biometric systems have been put into use for more than a century to identify and authenticate access to a secure environment. Biometric identifiers are related to the behavioral characteristics of a person. When compared to the traditional methods like token-based or knowledge-based authentication systems biometric based authentication systems are more reliable in verifying the identity of the person as biometric identifiers are unique to individuals.There are various issues associated with biometric authentication systems, the key issue is spoofing. The biometrics constitutes some of the features such as face, finger print, and so on. Presently, the face biometric are vulnerable to spoofing attacks. Examples for such type of spoofing attacks are presenting fake faces with mask, printed photos, videos etc . More commonly, the research carried out, includes six well known face recognition system namely, Face unlock, Veriface, Visidon, Facelock pro, Luxand Blink and and Fast access. These face recognition can be easily fooled without any difficulty and the only thing required to fool these systems are the images of the person to be spoofed. However, a valid person can login to his/her computer, office, ATM, bank account. The existing authentication methods are knowledge based (e.g. PIN and password). But, in some criteria a person can t login to the system at a situation he/she forget the secured password or someone tries to hack the password. It should be noted that, in real world applications, most of the biometric systems employed for securities purpose are uni-modal . Notably, for authentication process the uni-modal biometric system verifies only a single source of information. But this unimodal biometric system fails its capability to provide security, if some serious issues such as spoofing, non-universality, inter-class similarities, intra-class variations and noisy data enter into the systemThis thesis presents a detailed literature survey about fake face Identification and two techniques used for identif