Event-centric multisource stream processing and multimedia assimilation for geospatiotemporal phenomena

  • Authors:
  • Ashit Talukder

  • Affiliations:
  • Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, USA

  • Venue:
  • Proceedings of the 2nd ACM international workshop on Events in multimedia
  • Year:
  • 2010

Quantified Score

Hi-index 0.00

Visualization

Abstract

A new event-based multimedia stream data processing and representation solution is presented for large scale geographically distributed phenomena. Wide area geospatiotemporal patterns, such as cyclones or oil spills, have unique attributes and are captured in different media ranging from remote satellites, to in-situ sensors, to news video, photographs, and audio coverage. We pose an event-based mechanism to automatically detect such phenomena from raw untagged image streams, automatically extract various attributes for such events, establish correspondence of the automatically located event, and subsequently assimilate relevant information of the same event across the web from video, audio and static photographs. This disparate information from various sources can then be mashed up, customized for different communities, and disseminated to end-users. A use-case for automatic event-based cyclone notation and multimedia assimilation is presented.