University of Houston

In Academic / University environments lectures conducted by instructors are recorded and posted online so that students can benefit from watching these lectures if they missed them or want to listen to the whole / part of it again.

In this thesis I have worked towards implementing two algorithms which automatically finds indexes in a lecture video. The two algorithms are sequential and binary search. The sequential algorithm proceeds linearly through the video, comparing near consecutive frames to detect where slide transitions are happening. Binary search works its way by comparing end points of a video segment and then recursively calling itself till it narrows down to a frame. Both Binary and Sequential search provide with a list of slide transition points, which are large in number and not necessarily unique for every topic. In a second pass these transition points are filtered and index points are obtained. I also do extensive performance analysis of both these algorithms.