In the first step, the entire slide video is analyzed. We
try to capture every knowledge change between
adjacent frames, for which we established an analysis
interval of three seconds by taking both accuracy
and efficiency into account. This means that segments
with a duration smaller than three seconds
may be discarded in our system. Since there are very
few topic segments shorter than three seconds, this
setting is therefore not critical. Then we create canny
edge maps for adjacent frames and build the pixel
differential image from the edge maps. The CC analysis
is subsequently performed on this differential
image and the number of CCs is then used as a
threshold for the segmentation. A new segment is
captured if the number exceeds Ts1. Here we