Bernard Merialdo, Janne Saarela

Project description

We are investigating techniques for the automatic analysis of video sequences. Earlier work started with algorithms for cut detection, classification of shots,

We put a specific emphasis on the analysis of TV News programs. In particular, we are interested in techniques to create customized TV News programs by combining with Information Filtering approaches.

We are also studying the construction of audio-video summaries. We have formalized the problem of summarization as a problem of maximization with constraints. We have specified and investigated a number of constraints, and proposed a content-model for video segments.