Experiments on lecturer segmentation using texture classification and a 3D camera

Abstract

In our system for recording and transmitting lectures over the Internet the board content is sent as vector graphics, yielding a high quality image, while the video of the lecturer is sent as a separate stream. It is easy for the viewer to read the board, but the lecturer appears in a separate window. To eliminate this problem, we segment the lecturer from the video stream and paste his image on the board image at video stream rates. The lecturer can be dimmed by the remote viewer from opaque to semitransparent, or even transparent. This paper explains the two techniques we apply to achieve this: texture classification based segmentation, and segmentation using a novel 3D camera based on the time-of-flight of backscattered light principle. We argue that this technique provides a solution to the divided attention problem which arises when board and lecturer are transmitted in two different streams.