The automatic re-annotation of web images for improving access for blind users

Misman, Ahmad Fatzilah (2014) The automatic re-annotation of web images for improving access for blind users. PhD thesis, University of Nottingham.

[img] PDF - Repository staff only - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader
Download (3MB)

Abstract

The Internet has become the dominant means of communication in modern society. Increased use of and dependence on such things as email and the web has speeded up communication and allowed for a wide range of new ways in which we can communicate. The use of social media and Internet based shopping are amongst the most common forms of this development. Images are particularly common in such media. They are used for all sorts of purposes – from displaying products or pictures of individuals to decorations of all sorts. Blind people have the same rights and needs to use these modern technologies and so making sure that they can gain benefit from the images contained on these media is vital. The World Wide Web Consortium has drawn up specifications for the ways in which such images should be adaptable for best value by blind users. However, a large proportion of authors of web content do nothing to annotate the images they use in a way that would be of value to blind users.

A number of attempts have been made to create adaptation processes in which descriptions can be added to images after the page has been created. All of these methods use information and activity outside of the process of loading the pages – often involving large amounts of human effort. The work for this research has identified that it is possible to take the information that is already available inside the web pages themselves to enable valuable descriptions to be added to all image elements within pages in such a way that the usefulness of the images is improved for blind users. The software developed has made use of an algorithm which can define image purpose. In addition the algorithm uses mining of local data on the web pages to determine the relationship between the images and the text on the page to create an annotation of the images which is meaningful. An ethnographic study has revealed that the image descriptions should be clear, precise and meaningful and this has been the major aim of the software that has been developed. The mechanism was tested with two groups of blind users under different conditions. The first test revealed inadequacies in the software. These were tackled and the second test showed that the software enabled all users to gain better access to images in web pages that were adapted by the software than they were able to achieve for non-adapted web pages. A number of possible extensions to the work have been established.

Item Type: Thesis (University of Nottingham only) (PhD)
Supervisors: Blanchfield, P.
Brailsford, T.J.
Keywords: Web adapted image annotation, Image automatic descriptor
Subjects: T Technology > TA Engineering (General). Civil engineering (General)
Faculties/Schools: UK Campuses > Faculty of Science > School of Computer Science
Item ID: 14254
Depositing User: EP, Services
Date Deposited: 21 Jul 2015 14:49
Last Modified: 05 Jan 2018 23:04
URI: https://eprints.nottingham.ac.uk/id/eprint/14254

Actions (Archive Staff Only)

Edit View Edit View