Visitor Identification - Elaborating Real Time Face Recognition System

Georgy Kukharev, Adam Nowosielski
Technical University of Szczecin
Faculty of Computer Science and Information Technology
ul. Zolnierska 49
70-210 Szczecin

e-mail: {gkoukharev, anowosielski}



In this paper requirements and conditions for the visitor identification system are outlined and an example system is proposed. Two main subsystems: face detection and face recognition are described. Algorithm for face detection integrates skin-colour, mask analysis, facial features (fast and effective way of eyes localization is presented), reductors, knowledge and template matching. For face recognition a three stage algorithm is proposed. It utilizes well known methods connected in a sequential mode. To improve accuracy and speed some modifications to original methods were proposed and new one presented.

The aim was to build a visitor identification system which would be able to operate in mode with a camera and present results in real-time. The emphasis on speed and accuracy was stressed.