Visual block link analysis for image re-ranking

  • Authors:

  • Wengang Zhou;Qi Tian;Houqiang Li

  • Affiliations:

  • University of Science and Technology of China, Anhui Province, P. R. China;University of Texas at San Antonio, San Antonio, Texas;University of Science and Technology of China, Anhui Province, P. R. China

  • Venue:
  • Proceedings of the First International Conference on Internet Multimedia Computing and Service
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

Currently, image retrieval is becoming a popular feature in many search engines, most of which return images solely based on the metadata of pages in which the image appears. Since the page text may be inconsistent with the image content, image re-ranking is of great necessity. In this paper, we propose a Visual Block Rank approach for image re-ranking. We consider images as visual block sets and analyze the authority of images by exploring the underlying visual block link structure via LSA (Latent Semantic Analysis). The idea of Page-Rank is leveraged to our scheme for Visual Block Ranking and then the re-ranking order of images is deduced. We validate our approach with search results returned by Google Image. Experimental results show improvement of the accuracy and efficiency of our method over the state-of-the-art VisualRank algorithm.