Title: Employing physical distance in ranking image-search results

Authors: Natheer Khasawneh; Rami A. Al-Hailat; Ahmad T. Al-Hammouri

Addresses: Department of Software Engineering, Jordan University of Science and Technology, Irbid 22110, Jordan ' Souq.com, 8 Nuh Al-Rumi Street, Sweifiyeh, Amman, Jordan ' Department of Network Engineering and Security, Jordan University of Science and Technology, Irbid 22110, Jordan

Abstract: In this paper, we enhance web image searching by exploiting the physical distance between an image on the web page and the surrounding text. The rationale is that the closer a term word to an image is, the more strong the relation between that word and image is. The physical distance is measured between the exact positions of a term and an image inside a webpage as it appears when it is completely rendered by a web browser. We show that incorporating physical distance in the ranking process indeed reduces the number of irrelevant images in the returned search results.

Keywords: physical distance; image search; image retrieval; search results ranking; scoring; irrelevant results; visual representation; web images; term words; surrounding text.

DOI: 10.1504/IJIIDS.2014.068341

International Journal of Intelligent Information and Database Systems, 2014 Vol.8 No.4, pp.324 - 339

Received: 11 Jul 2013
Accepted: 21 Jan 2014
Published online: 10 Apr 2015 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article