Intelligent Tracking Teaching System based on monocular active vision

Abstract

Teacher detection and tracking is the most important and fundamental functionality in the implementation of the Intelligent Tracking Teaching System (ITTS). In order to track teacher's movement in real-time, face tracking in rostrum region is initiated by the normalized size face adaboost detection followed with the Expectation Maximization (EM) algorithm based on HSV color space and prediction of face position. The split line and position-based visual servo were adopted to realize the tracking strategy, which is trying to keep the teacher in the middle of image by controlling pan/tilt/zoom monocular camera in either rostrum region or classroom region. Furthermore, the student camera will adaptive pan/tilt during the interaction process between teacher and students, and a real-time display on the GUI window is given. The experiment results demonstrated fast and extremely smooth pursuit for teacher's motion despite time-varying variation in illumination, with sustained robustness to the change of pose (e.g., from frontal face to nearly back of head).