New Tools for Automating Tactile Geographic Map Translation
Nizar Bouhlel, Anis Rojbi · 2014 · Proceedings of the 16th International ACM SIGACCESS Conference on Computers & Accessibility (ASSETS) · doi:10.1145/2661334.2661335
Summary
This demo paper presents a Matlab-based software tool that semi-automates the conversion of visual geographic maps into tactile versions suitable for blind students. Tactile graphics are the most effective modality for blind people to comprehend graphical images, but converting visual maps to tactile form traditionally requires tactile graphic specialists (TGS) to perform non-trivial, time-consuming, expensive manual steps. The software integrates a seven-step pipeline into a single application: (1) image acquisition and pre-processing (scanning and noise filtering), (2) image segmentation (partitioning the map into objects or classes using Fuzzy C-Means clustering, with the number of classes set by the user or TGS), (3) image simplification (reducing detail and complexity using image processing techniques), (4) texture attribution (assigning distinct tactile textures to each colour or class of the segmented image — a key feature, as texture is rapidly grasped through touch due to spatial layout), (5) optical character recognition (OCR) to identify text labels on the map, (6) Braille translation of the recognised text, and (7) tactile image output. The two main differentiators from prior work are: integrating all tools into a single application useful to the community (rather than disparate separate tools), and the automated texture attribution for segmented image regions. After clustering, classified text can be removed and filled by the surrounding region's colour, with text labels converted to Braille and repositioned. The texture assignment allows users to translate the image to tactile form in ways analogous to current manual practices, but with significantly reduced effort.
Key findings
The software successfully integrates image processing, machine learning-based segmentation, OCR, and Braille translation into a unified pipeline for tactile map production. The Fuzzy C-Means clustering algorithm effectively segments geographic maps into distinct regions (countries, bodies of water, terrain types) that can each be assigned unique tactile textures. The OCR component addresses a challenge specific to map translation: visual maps contain text labels (country names, city names, geographic features) that must be identified, removed from the visual layer, and replaced with Braille equivalents in the tactile version. The semi-automated approach retains human oversight — the TGS can set the number of segmentation classes and review results — while eliminating the most tedious manual steps. The complete pipeline within a single application is a practical advance over previous approaches that required TGS to use multiple separate tools and manually transfer data between processing steps.
Relevance
Geographic maps are among the most common graphical materials in education, and their inaccessibility represents a significant barrier for blind students studying geography, history, and social studies. Producing tactile maps manually is so labour-intensive that many blind students simply never have access to maps that their sighted peers use routinely. For accessibility practitioners working in education, this tool addresses a practical production bottleneck — if tactile maps can be produced more quickly and cheaply, more blind students can have access to geographic learning materials. The texture attribution step is particularly important because it directly addresses how blind people read tactile graphics: distinct textures allow rapid differentiation of map regions through touch, in the same way that colours differentiate regions visually. The integration of OCR and Braille translation automates one of the most tedious aspects of map conversion — manually transcribing every text label. While this is a brief demo without formal evaluation, the approach of semi-automating the production pipeline for tactile educational materials has continued to develop as image processing and AI capabilities have improved.
Tags: tactile graphics · blindness · braille · image processing · OCR · accessible maps · education · geographic maps