Multimedia Management Project

We were one of the first to develop a theoretical model of multimedia data that handles images, video, text as well as relational data. Our current work focuses on the following.

• VIDEO ALGEBRA: We are implementing the AVE! system (Algebraic Video Environment). AVE! includes an extension of the relational algebra to query video sources. It includes a mix of image processing methods to extract video content (provided by our colleagues at the University of Naples, Italy) and annotation methods. We are currently finishing the implementation of AVE!.
• Video Summarization: There are many applications where summaries of videos are needed, rather than the video as a whole. For instance, in a surveillance application, video summaries may show critical events. Likewise, in a summary of a colloquium talk, we may show the main contribution slide. We are developing a video summarization system called CPR which is based on three criteria that a summary must satisfy (Continuity of the summary, Priority objects should be included, Repetition must be avoided).
• PowerPoint Databases: We have developed a formal model of PowerPoint databases that allows the automatic extraction of information from PowerPoint sources and supports their storage in a database. We have extended the relational algebra to handle querying of PowerPoint sources and developed a set of equivalence results. The PowerPoint algebra has been fully implemented on top of Oracle.