Line Graph Interpretation Tool for VisuallyImpaired Users

Read full text Summarise with VIRAT
Authors: Ajitesh Parihar, Manpreet Kour, Mohamed Saber
Cited by: 0
References: 5

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Line diagrams and mathematical function plots arecommonly used in scientific and educational textbooks andarticles to convey quantitative relationships. However, their visualnature presents significant challenges and barriers for visuallyimpaired learners. Despite the ongoing efforts to improve acces-sibility in higher education, the non-text content still remainsdifficult to interpret non-visually. The existing diagram analysisapproaches often focus on chart type classification, requiremanual user intervention for tasks such as label mapping,or fail to provide accessible end-to-end interaction for nonvisual analysis. Also, many of these systems do not explicitlyconsider the requirements of visually impaired users, resulting inworkflows and interfaces that are not accessible. In this paper, wepropose an accessibility aware framework for automated analysisof line diagrams from raster images. The proposed end-to-endsystem performs CNN based curve segmentation, diagram typeclassification, label and tick extraction using optical characterrecognition, mapping of curve pixels to the diagram domain, andsupports interactive querying along with a user interface com-pliant with the Web Content Accessibility Guidelines. Evaluationis conducted using a large synthetic dataset of 2D chart imagesrepresenting a variety of mathematical functions and discretepoint line charts, which is used for both model training andsystem-level assessment

Fields of Study & Subjects

Digital Accessibility for Disabilities Handwritten Text Recognition Techniques Data Visualization and Analytics

Publication details

Journal / source
HAL
Publication date
24 March 2026
Language
EN
Document type
Preprint