Instant Tactile-Audio Map: Enabling Access to Digital Maps for People with Visual Impairment
Zheshen Wang, Baoxin Li, Terri Hedgpeth, Teresa Haven · 2009 · Proceedings of the 11th International ACM SIGACCESS Conference on Computers and Accessibility (Assets '09) · doi:10.1145/1639642.1639652
Summary
This paper presents an automated system for converting standard digital map images into interactive tactile-audio maps that people who are blind can explore independently. The core problem addressed is that while online map services like Google Maps provide rich geographical information to sighted users, people with visual impairments cannot access the map images themselves — they may get textual directions but miss the spatial layout of roads, landmarks, and neighbourhoods. Existing methods for creating tactile maps require a sighted professional to manually trace and simplify graphics, a process that is time-consuming and labour-intensive, preventing real-time access. The proposed system automates this conversion through a pipeline with two main components. First, a text detection and segmentation module uses spatial frequency analysis, morphological thinning, Hough transforms, and OCR to extract, orient, and recognise text labels from the map image. Second, a graphics recreation module simplifies the remaining visual elements into a tactile-friendly form through contrast enhancement, thresholding, noise filtering, and insertion of tactile symbols (raised or sunken dots) at interactive text regions. The system outputs both a tactile hardcopy (printed via a tactile embosser or swell paper) and an SVG file. When a user places the tactile printout on a touchpad, touching raised symbols triggers audio playback of the corresponding text labels, creating an integrated tactile-audio exploration experience.
Key findings
The text extraction module achieved approximately 100% detection accuracy and 90% recognition accuracy across a dataset of 23 maps containing 549 text labels. Six blind volunteers evaluated the system through navigation tasks on five maps of varying complexity, achieving an average landmark location accuracy of 0.93 (out of 1) and route tracing accuracy of 0.87. A follow-up user experience survey yielded an overall average score of 8.0 out of 10 (8.5 excluding the first participant, whose results were affected by a calibration issue). Participants found the interactive tactile-audio maps informative, easy to use, and helpful for walking navigation. The system processes a map in approximately 20 seconds on a standard PC, making real-time access feasible. Key challenges identified included users needing both hands to trace routes while holding landmarks, confusion from discontinuous street segments with the same name, and the need for a calibration step between the tactile printout and the touchpad.
Relevance
This research addresses a significant gap in map accessibility by automating a process that traditionally required extensive manual effort from sighted professionals. The practical implication is that a person who is blind could take any digital map image and independently produce an explorable tactile-audio version in seconds rather than days. The multimodal approach — combining tactile spatial information with audio text labels — demonstrates how different sensory channels can complement each other to convey complex spatial information. While the prototype had limitations in handling very complex maps and low-resolution inputs, the user evaluation confirmed real-world utility for local navigation planning. The system components (text detection, graphics simplification, SVG generation) could also be applied more broadly to making other types of visual information accessible, such as floor plans, transit maps, or data visualizations.
Tags: tactile maps · visual impairment · multimodal interaction · audio feedback · navigation · computer vision · OCR · tactile graphics