Face Recognition Using Principal Component Analysis Applied to an Egyptian Face Database

Although face recognition is highly race-oriented, to-date there is no Egyptian database of face images for research purposes. This paper serves two purposes. First we present the efforts undertaken to build the first Egyptian face database (over 1100 images). Second we present a variant algorithm based on principal component analysis (PCA) but adjusted to Egyptian environment. In order to conduct face recognition research under realistic circumstances, no restrictions have been imposed on the volunteers (eyeglasses, moustaches, beards, and veils (hijab)). Furthermore, photos, for each volunteer, were taken during two sessions that are two months apart (March and May). Meanwhile, multiple light sources have been used. More than 1000 experiments have been carried out to evaluate the approach under different conditions. A new pentagon-shaped mask has been devised, which has proven suitable to enhance the recognition rate.