Graphical information for blind users of the Web

Yanxia Yang, Purdue University

Abstract

New information technology offers great potential to improve access to information by blind and visually impaired people. This study investigates the potential of using computer automated image processing to improve blind users' understanding and accessibility to information in graphical format on the Web. Based on a review of the literature, three hypotheses are presented. First, when blind users obtain information from a Web page, they will better comprehend images with lower complexity. Second, blind users will find computer preprocessed graphic images more understandable than the original images. Blind users will have the best understanding of symbol images simplified by the edge detection method. Third, providing blind users simplified graphical information when they browse Web pages will improve their performance compared with unsimplified graphics. Two experiments were conducted to test these hypotheses. The experiment results support Hypothesis One and Hypothesis Three, and partially support Hypothesis Two. Conclusions drawn from these experiments are as follows: (1) After simplification, blind users find it easier to understand Web photographic images. The segmentation method was found to be a little more effective than the edge detection method as a means of simplifying Web photographic images for blind users; (2) The simplification process was less effective for less complicated Web images, such as symbols. Either the segmentation method or the edge detection method showed similar effects; (3) Providing graphical screen layouts of Web pages simplified using the segmentation method resulted in improved performance of blind users on Web browsing tasks. Recommendations for image identification, subject recruiting, image selection, and simplification method are made for future studies.

Degree

Ph.D.

Advisors

Lehto, Purdue University.

Subject Area

Industrial engineering

COinS