Presenter tracking in a classroom environment

摘要

As intelligent environments become more focused on supporting human activity, we see a growing need for the computer to ascertain users' location and activity, be it for face detection, gesture recognition, or even simple tracking. To support such applications, a robust tracking algorithm is required. A camera tracking algorithm for a pan and tilt camera is proposed, that is robust to moderately dynamic backgrounds and varying lighting conditions, while requiring only modest computational resources. The algorithm is based on pre-filtering as well as temporal differencing, and achieves impressive results in the real-world setting of a university classroom.