A study on face tracking in real-time for robot

Abstract

This paper proposed the face detection and tracking system using a embedded computing system for humanoid robot environment which allows the user to observe of recognize an unknown person. For detection faces in image sequence, the system uses the skin color model and deformable ellipse matching. Faces in a view are detected by maximizing the image gradient magnitude around the perimeter of the ellipse (or face line) and the number of skin color pixels in the ellipse area. The information of detected face regions are stored on-line into a face state table and are evaluated for tracking in real-time. Results of this processing are applied to embedded humanoid robot The applied robot acts to detect human's fate and react to it. After this processing, the result is sent to embedded humanoid robot through zigbee wireless module. The robot that got this data acts to detect human's face and react to it.