Design and implementation of Pan-Tilt control for face tracking

Abstract

This paper presents design and implementation of face tracking system using servo-motor-controlled Pan-Tilt visual system. The objective is to maintain the position of a tracked object at the center of frame window on the screen. A simple camera is used as visual sensor to obtain the face object's actual position. Combination of Viola-Jones and Template Matching algorithms are used for face detection, in which template matching method acts as a back up on the occasion Viola-Jones fails to detect the face object. Object's position on the screen is controlled indirectly through the position of the camera using an Arduino-based controller. Two control methods are implemented to track the object, namely lead-lag compensator and PID control. To obtain proper parameters of the controller, identification of the Pan and Tilt system has been done and validated. Both controller algorithms are then compared and analyzed in accordance with system design specification. Overall, experimental results show that PID has faster transient response than lead-lag either for Pan or Tilt.