A vital SVKPCA feature set for robust FRS with ensemble neural network classifier

Abstract

Face is the robust biometric in the field of access control and recognition. In this paper FFT Set, HARA Set, FHA Set, FHAKP Set and SVKPCA set are the five facial feature sets which was formed from spatial and frequency domain are analyzed using ensemble Neural Network to design a robust FRS. The ORL, NIR and Indian face databases are used to perform the experiments to prove that the proposed singleton SVKPCA set gives promising results irrespective of many challenges existing in the face databases. Following are the challenges faced by the feature set: Gender, pose, expressions, scale and timing. The Neural classifier used in this proposed work incorporates the ensemble approaches of bagging and boosting to enhance the accuracy of the FRS from its regular standard model.