Converting SVG Images to Text and Speech

There are contents with a strong visual component, e.g., in the engineering field, such as technical drawings, charts, diagrams, etc., virtually inaccessible for people with visual disabilities (blind, visually impaired, etc.). These contents are mostly vector based and can generally be found on the Web in various formats, but the recommended one is SVG. The authors created an online application to convert SVG images (containing simple geometric figures of various sizes, filling colours and thickness and colour filling contours) in textual and spoken description by a speech synthesizer, client side based, without browser plugins. This application also allows the user to navigate with efficiency through the image description using four levels of detail and keyboard commands. In this paper, the authors propose a novel method for image description based on the Gestalt theory, considering the cognitive load and, for the first time, providing users with visual impairments access to the full content of SVG images. Application tests were carried out with 11 users (eight normally sighted, two blind and one amblyopic), comparing descriptions made by the application and by humans. The authors concluded that, for all users, there were improvements of 9%, using the application. Considering only the visually impaired, this figure rises to 18%.