Home // MMEDIA 2011, The Third International Conferences on Advances in Multimedia // View article

A Clustering-based Approach to Web Image Context Extraction

Authors:
Sadet Alcic
Stefan Conrad

Keywords: Image Context Extraction, Web Content Mining

Abstract:
Images on the Web come along with textual descriptions that are valuable for different applications, such as image annotation, clustering of images, image categorization, etc. But usually Web pages are poorly structured and cluttered with contents of different topics, which hinder the accurate detection of the image context. Existing approaches are based on heuristic rules and thus cannot handle the variety of documents on the Web. In this paper, we introduce a novel approach to image context extraction, building on a Web content distance measure. Utilizing this distance measure, the addressed problem can be reduced to a content clustering problem where an image is associated with the textual contents of the cluster it belongs to. Our evaluation studies confirm the validity and quality of the proposed method and demonstrate its applicability to the Web.

Pages: 74 to 79

Copyright: Copyright (c) IARIA, 2011

Publication date: April 17, 2011

Published in: conference

ISSN: 2308-4448

ISBN: 978-1-61208-129-8

Location: Budapest, Hungary

Dates: from April 17, 2011 to April 22, 2011