Comparison of several techniques for detection of key slides in lecture support materials

In this paper we deal with the two tasks i) with the detection and removal of duplicate images and ii) with the detection of key-slides in lecture support materials. The keyslides can provide an overview of the lecture as well as they allow us to split the lecture into content-homogenous segments. The duplicate image removal spares resources when storing the materials. Moreover when streaming the lectures to device with mobile internet access the spared transmission capacity becomes important. We propose and compare three different approaches to solve the above mentioned tasks. The first approach uses color histograms and line-profiles for the image parameterization. The second one utilizes the video-encoder key-frame detection. The third approach is based on motion vectors analysis. The experimental results show that our approach is reliable enough for a real deployment.