Skip to Main Content
With the recent advances in consumer electronics and the need for noninvasive biometric recognition systems, face recognition has become a viable option at least computationally. The aim of this paper is to present a complete face recognition system that works in real time, and is robust to illumination and expression for the most part. The system was developed to study the feasibility of employing facial recognition in relatively controlled environments like that of an ATM machine. The system employs a statistical skin color model to detect faces and then extracts them for eye detection that works within the framework of EBGM (Elastic Bunch Graph Matching) algorithm. Finally, face graphs are extracted and matched to those within a database. The final prototype was tested using a 1.3 megapixel USB webcam and a Pentium 4 2.4 GHz processor.