Kinect for video conference

Abstract

Like in any particular conversation, one can tell your interest and engagement via your facial expressions. In current video conferencing systems involving webcams embedded into your personal computers, effective communication is often hindered by the lack of the eye contact. While it is true that this problem has been partially solved for high-end video conference systems using expensive equipment, for the everyday consumers, a cheap and viable solution is still not within our grasp. Many works out there have tried to solve the eye contact problem. However, they involve highly technical algorithms and the use of several external devices. My proposed method involves the use of a single Kinect device to help solve this problem. The approach I took towards this was basically a combination of Kinect SDKs, OpenGL and OpenCV. From the Kinect data streams, we will extract and draw out the point cloud using OpenGL. We will also apply tilting in according to correct the gaze of the user. This will portray the illusion that the user is making eye contact when he is video conferencing even if he is not.