Thursday, January 22, Volen 101, 4:10-5:10 pm. (Refreshments at 4:00pm)
This talk will present multiple media stream correlation and cross-modal information retrieval. Multiple media stream correlation provides a framework for locating and exploiting correlations between multiple, potentially heterogeneous, media streams. The goal is to determine temporal and spatial alignments which optimize a correlation function and indicate commonality and synchronization between media streams, providing a basis for comparison of media in unrelated domains. There are many real-world applications for this technology, including language translation alignment, speaker localization, and degraded media realignment. The specific application of text-to-speech synchronization will be described as an example. Cross-modal information retrieval exploits this computed synchronization in order to support queries in one media with results in another. The structure of the Xtrieve cross-modal information retrieval system, currently under development at the Dartmouth Experimental Visualization Laboratory, is described as an example of this technology.
Host: Jim Storer