Metadata-based automatic image tagging
by Randi Karlsen; Martin Hætta Evertsen; Najeeb Elahi
International Journal of Metadata, Semantics and Ontologies (IJMSO), Vol. 8, No. 4, 2013

Abstract: Today, keyword-based search is the most common technique for searching images, and requires the availability of tags and annotations that reflect image content. While there are many techniques for automatic annotation based on image content analysis, less work has utilised image metadata as a basis for image tagging. We investigate the ability to automatically tag images based solely on image metadata, and present a novel approach to image tagging using a combination of the metadata geo-location, date/time and category keyword. We aim to collect relevant tags for a target image by selecting tags from related images on a community image collection, such as Flickr. Contributions of our approach are the use of dynamic techniques that adapt tag selection to the availability of relevant images and tags in the community collection. Through testing we demonstrate the usefulness of the techniques and the use of metadata information in the image tagging process.

Online publication date: Sun, 22-Dec-2013

The full text of this article is only available to individual subscribers or to users at subscribing institutions.

Existing subscribers:
Go to Inderscience Online Journals to access the Full Text of this article.

Pay per view:
If you are not a subscriber and you just want to read the full contents of this article, buy online access here.

Complimentary Subscribers, Editors or Members of the Editorial Board of the International Journal of Metadata, Semantics and Ontologies (IJMSO):
Login with your Inderscience username and password:

    Username:        Password:         

Forgotten your password?

Want to subscribe?
A subscription gives you complete access to all articles in the current issue, as well as to all articles in the previous three years (where applicable). See our Orders page to subscribe.

If you still need assistance, please email