A Hybrid Region Weighting Approach for Relevance Feedback in Region-Based Image Search on the Web

  • Authors:
  • Deok-Hwan Kim;Jae-Won Song;Ju-Hong Lee

  • Affiliations:
  • School of Electronics and Electrical Engineering, Inha University, 253 Yonghyun-dong, Nam-gu, Incheon 402-751, Korea;Department of Computer Science and Information Engineering, Inha University, 253 Yonghyun-dong, Nam-gu, Incheon 402-751, Korea;Department of Computer Science and Information Engineering, Inha University, 253 Yonghyun-dong, Nam-gu, Incheon 402-751, Korea

  • Venue:
  • SOFSEM '07 Proceedings of the 33rd conference on Current Trends in Theory and Practice of Computer Science
  • Year:
  • 2007

Quantified Score

Hi-index 0.00

Visualization

Abstract

In this paper, a new hybrid weighting method, which learns region importance from the region size and the spatial location of regions in an image, is introduced to re-weight regions optimally and improve the performance of the region-based search system on the Web. Relevant images marked by an user may exhibit very different visual characteristics so that they may be scattered in several clusters in the feature space, since there exists the semantic gap between the low level feature and the high level semantics in user's mind. Our main goal is to find semantically related clusters and their weights to narrow down this semantic gap. To do this, The hybrid region weighting method, which refines the weights of region-clusters through relevance feedback, determines the importance of regions according to the region size and spatial location information of regions in an image. Experimental results demonstrate the efficiency and the effectiveness of the proposed weighting method in comparison with the area percentage method and the region frequency weighted by inverse image frequency method, respectively.