SOFTWARE TOOL FOR WEB CONTENT ACCESSIBILITY ANALYSIS FOR VISUALLY IMPAIRED USERS

Abstract

The article addresses the issue of implementing modern principles of digital inclusivity in web content. It is shown that a key factor affecting the accessibility of a web resource for people with visual impairments is the contrast level of its web pages and their content, including elements of the graphical user interface. An original approach is proposed for automated contrast analysis of web content in accordance with the Web Content Accessibility Guidelines (WCAG) 2.1 standard, without accessing the Document Object Model (DOM) structure or HTML code of the web page, using only its graphical representation. To implement this approach, the following popular tools were used: programming language Python, OpenCV, and Tesseract Optical Character Recognition. To improve the performance of the developed software, principles of parallel computing were implemented at its core to enable distributed processing of designated text blocks across multiple threads. For this purpose, the concurrent.futures library was used, allowing the creation of threads that process different parts of the image – where text has been identified – in parallel. The developed module demonstrated its ability to effectively analyze the contrast of web page elements without access to their DOM structure, making it suitable for automated accessibility audits of images, PDF documents, web interfaces, and other graphical elements. This was confirmed during testing of the module on several real websites, such as the Medium platform (access mode: https:// medium.com/). Comparative testing was also conducted between the developed module and well-known alternatives, including Google Vision API, Adobe Acrobat OCR, and Tesseract with custom scripts. The results were analyzed based on a dataset consisting of various types of images of different sizes, according to the following criteria: image processing time, accuracy of text block and contrast detection, and operational stability.