Automatic framework for efficient pose invariant face recognition using 2d pose normalization

Abstract

In recent years, Biometrics has made a significant impact on society. Biometric-based verification, authentication is gaining immense popularity in the digital world, since it is helping people to get rid of the possession of the traditional identity such as ID cards, documents, tokens, PINs, Passwords and enabling the person to use their unique physiological or behavioural characteristics. Today, the biometric applications are used across institutions, industries, government establishments and in many business outfits. Biometric modalities can be classified into three categories such as Physiological (fingerprints, hand geometry, facial recognition, retina, DNA etc.), Behavioral (signatures, typing rhythm, gait etc.) and a combination of these two (voice). Deploying a biometric modality depends on the authentication or identification application. In recent days, a popular biometric physiological modality of facial structure is used for various applications of authentication and identification of individuals in an indoor, outdoor surveillance, public monitoring environments such as airport terminals, railway stations, bus terminals, commercial areas etc., because of its unique advantages compared to the other biometric traits. Face recognition technology is widely preferred in various applications because of its non-intrusive, non-contact nature and high level of acceptability. Despite rapid development in this technology, satisfactory recognition accuracy is achieved for the faces captured in the controlled, semi-controlled environment with constrained settings because the faces are captured with the user cooperation newline