Browse Prior Art Database

Assistive Technology Enhancement for Image Captions

IP.com Disclosure Number: IPCOM000240881D
Publication Date: 2015-Mar-09
Document File: 2 page(s) / 22K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed is a method that utilizes existing mature image searching technology to associate an image with no caption information with other similar images or graphs that contain caption information, and then derive valid alternative text for the first image.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 52% of the total text.

Page 01 of 2

Assistive Technology Enhancement for Image Captions

Visually impaired users are unable to use Assistive Technology devices (e.g., screen readers) in order to get information about images unless the images have without associated text.

A method is herein disclosed for utilizing existing image searching technology to associate an image with no caption information with other similar images or graphs that contain caption information, and then derive valid alternative text for the first image. This allows visually impaired users to receive information about the image when the caption is not immediately available from metadata.

The method uses image search technology to conduct image searches. If found images are somewhat matched, and if an image in the search results has a caption, then the method records that caption.

For example, Image X is on a website without a caption. The method finds a similar image on another website, and that image has a caption (i.e. "A Figure of X Personalized"). The method applies the caption of "A Figure of X" to the first image.

The method is also capable of scanning the text on the page that surrounds the image, assuming that it is relevant to the image and storing this text. The method then compares the text collected from both a second image (found as the result of a search) and the text found on the web page. The intersection of these two groups of text comprises the results of the caption of the image.

If the disclosed method may not identify an intersection of two groups of data, then the method further expands the search terms by using the synonyms of the text adjacent to the image. Using the example above, the method may search for synonyms of the

word, "personalized".

If the method still returns no results, then it may use the definitions of the text adjacent to the image to conduct a search to identify whether any returned strings overlap with the captions from the image search results.

For example, assume that in the right column adjacent to the figure, "A Figure of X Personalized," is an image labeled, "Sign in and personalize." The method:

1. Performs a search for the term, "personalize"

2. Finds nothing in the search results identified as "personalize"

3. Expands the search...