Indexing and retrieval of interpreted video contents

The work will focus on the research of indexing, retrieval and visualization schemes for searches over large video pools. The main goals of this research could be summarized as the following:

• new feature extraction methods as the base for higher level interpretation,
• new indexing structures and algorithms (theoretical and proof-of-concept) supporting fast multi-feature queries,
• exploit results in object segmentation and extraction,
• methods for assigning automatic annotations based on the extracted features,
• new visualization methods of retrieved results,
• based on the above, a video retrieval framework based on content interpretation.

The work will produce new theoretical and practical results in semantic video analysis and content-based retrieval (indexing, interpretation, retrieval, visualization).