Face recognition in a heterogeneous population

Face recognition technology has many and varied applications, ranging from document verification to crowd surveillance. The use of face recognition technology in access control is, due to its nonintrusive nature, extremely promising. Most face location and segmentation algorithms proposed in the literature have very limited capabilities, with most not able to find a face which is either rotated or scaled. A novel segmentation algorithm which attempts to overcome these restrictions was developed and used in this study. The Eigenface approach (Karhunen-Loeve transform) has been shown to work extremely well when used to extract feature vectors from a face image. It was found to be the most promising of all the published methods and was therefore used. Most face recognition papers in the open literature have generally used Caucasian faces to test their algorithms, no special mention is made of the other population groups. The effect of a heterogeneous population on face recognition is therefore something of an unknown quantity. It is the aim of this paper to investigate the performance of the developed face recognition system, with emphasis on the effect of a heterogeneous population