Recognition performance analysis of Subpattern-based Principal Component Analysis for different image partition dimensions and different preprocessing methods

Principal Component Analysis (PCA), as one of the most used method in face recognition application, is an analysis method aimed at representation of the multivariate data structural. The PCA method, is a linear transformation which maps the high correlated multivariate data to a new coordinate system where the data is uncorrelated. In this paper as a kind of the traditional PCA method called Subpattern-based PCA method's recognition performance is evaluated under different partition dimensions of face images and for different preprocessing methods. In the experimental studies ORL is used as database.