CONTENT-BASED BROWSING IN LARGE NEWS VIDEO DATABASES

In this paper, we have evaluated the effectiveness of novel content-based browsing paradigm in video retrieval and compared it to the traditional model of content-based querying with relevance feedback. The presented model, cluster-temporal browsing, integrates feature clusters and chronological video structure dynamically in a single browsing view. A prototype of the browser has been evaluated in 70 hour news video collection by 17 test users in 24 predefined and highly semantic search topics. The contributions of this paper are: comparison of clustertemporal browser against prevalent content-based query paradigm and evaluation of the effect of various browser parameters such as features and interface configurations in search performance. The main conclusions of the paper are that the cluster-temporal browsing surmounts traditional search paradigm in semantic search topics and that lowlevel visual features do not add to the search performance that is obtained using text search on speech transcripts.