Mixture model based contextual image retrieval

  • Authors:
  • Xing Xing;Yi Zhang;Bo Gong

  • Affiliations:
  • University of California Santa Cruz, Santa Cruz, CA;University of California Santa Cruz, Santa Cruz, CA;Oracle Corporation, Redwood Shores, CA

  • Venue:
  • Proceedings of the ACM International Conference on Image and Video Retrieval
  • Year:
  • 2010

Quantified Score

Hi-index 0.00

Visualization

Abstract

Traditional image retrieval techniques search for images without considering the query context. When they are applied to applications like annotating text with images, the results might not be satisfactory because the meaning of a word/phrase varies depending on the context. In this paper, we propose an alternative method to retrieve images based not only on the query but also on the query context. We exploit the text surrounding a query in a document and the disambiguated sense of the query as the query context. We also propose a mixture model to retrieve the context dependent images. Experimental results show that significantly better retrieval performance can be achieved when query context is taken into account.