Multi-Modal Facial Recognition Based on Improved Principle Component Analysis with Eigen Vector Feature Selection

Physiological biometric behavior such as face is taken in our research work to verify the individual’s identity. The process involved in biometric system is feature detection and recognition process. However, face recognition is hard case on the biometric object recognition with higher performance result. Most of the existing face recognition system takes longer time to produce the performance result, due to the more dimensionality distractions. For better recognition accuracy, new face feature extractions are needed to be explored. To develop an effective face recognition system, Multi-modality Face Feature Recognition System based on the Principle Component Analysis Eigenvector Selection (PCAES) Method is proposed in this paper. Initially, Invariant Face Region Feature Detection operation extracts the face from the total training image of the users. Secondly, the new face features are selected using the Eigen vector selection through branch and bound process. For the effective face classification process in PCAES Method, new logistic regression learning classification process predicting the outcome of a face categorical dependent variable with lesser dimensions. Finally, Principle Component Procedure with R3 rule procedure increases the recognition level accuracy. PCAES finds the new training multi-modality data representation in the subspace of smaller size which maximizes the dimensionality reduction level. The reduction of dimensionality reduces the distraction level. Case study and preliminary experimental results conducted in viable PCAES approach on proving the 2D and 3D facial recognition. Experiment is conducted on factors such as, feature extraction indexing rate, false acceptance rate, false rejection rate and recognition accuracy level.