An intensity and size invariant real time face recognition approach

Abstract

This paper proposes an intensity and size invariant real time computer vision-based face recognition approach. With this method, human facial area(s) are first detected automatically from real-time captured images. The images are then normalized using histogram equalization and contrast stretching. Finally face is recognized using eigenfaces method. This proposed method is camera to face distance invariant as well as intensity invariant. The effectiveness of this method was demonstrated using both static and dynamic images. The average precision and recall rate achieved by the proposed method is above 90%. © 2009 Springer Berlin Heidelberg.