A Spoofing Security Approach for Facial Biometric Data Authentication in Unconstraint Environment

Security is ever a challenging issue of research at national and international level due to the privacy standards of object detection and recognition in unconstraint environment. Almost real-life activities are performed under the unconstraint conditions in which exact determination of any activity fails due to the lack of complete information of facial parts, hands and relevant objects. However, from social interaction views, face detection is a quite saturated research in normal conditions but, due to highly sensitive and easy availability of facial data, encourages the researchers to work by cryptographic aspects in unconstraint environment. In this work, we focus on securities issues for automated facial biometric data authentication by local features extraction. By keeping space and time intricacy, we ensure the fusion of Gabor, center-symmetric LBP and discriminative robust LBP features to improve the performance. The feature matching is performed by majority of vote in which difference of Gaussian and robust local ternary pattern is used. Since we choose one sample to match with stored faces, the reduction in space complexity improves the performance up to 89% of matching accuracy.