Abstract: Security systems based on traditional techniques are more prone to be attacked by hackers and, such major attacks exposed weakness in most sophisticated security systems. A number of supervision agencies are now more motivated to improve security based on body or behavioral uniqueness, often called biometrics. Face recognition is non-intrusive since it is based on images recorded by a distant camera and can be very effective even if the user is not aware of the existence of the face recognition system. Automatic face recognition is a challenging problem in the field of Pattern Recognition and image processing due to its varying nature. It has wide range of applications such as law enforcement, national identity, banking, and logical access control. In this research paper a modification of an existing technique (PCA) will be discussed to improve the efficiency of face recognition. In Principal Component Analysis (PCA) Technique face images are represented as vectors by concatenating the pixels of the image line-by-line. A mean face is calculated by computing the average of each vector. A difference vector is also computed for each user to qualify the differences to the mean face. Than a covariance matrix of the difference vectors is computed. As a final step, principal axes are obtained by Eigen decomposition of covariance matrix. First N eigenvectors presenting the highest Eigen values will be retained and represents the most significant features of face images. Finally, each user model is represented as a linear combination (weighted sum) of coefficients corresponding to each Eigen faces. PCA is performed only for training the system, due to which this method results to be very fast when testing new face images. Due to this property PCA is selected for this work as modification will take place during training phase.

Keywords: AFR, PCA, Face Recognition, Genetic Algorithm, Age Invariant.