Scene signatures for unconstrained news video stories

  • Authors:

  • Ehsan Younessian;Deepu Rajan

  • Affiliations:

  • School of Computer Engineering, Nanyang Technological University, Singapore;School of Computer Engineering, Nanyang Technological University, Singapore

  • Venue:
  • MMM'12 Proceedings of the 18th international conference on Advances in Multimedia Modeling
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

We propose a novel video signature called scene signature which is defined as a collection of SIFT descriptors. A scene signature represents the visual cues from a video scene in a compact and comprehensive manner. We detect Near Duplicate Keyframe clusters within a news story and then for each of them we generate an initial scene signature including most informative mutual and distinctive visual cues. Compared to conventional keypoint-trajectory-based signatures, we take the co-occurrence of SIFT keypoints into account. Moreover, we utilize keypoints describing novel visual clues in the scene. Next, through three steps of refinements on the initial scene signature we shorten the semantic gap to obtain more compact and semantically meaningful scene signatures. The experimental results confirm the efficiency, robustness and uniqueness of our proposed scene signature compared to other global and local video signatures.