Understanding How Low Vision People Read using Eye Tracking

Title of the Paper

Understanding How Low Vision People Read Using Eye Tracking

Paper Information

• Subject Areas: Assistive Technology, Low Vision Research, Eye Tracking
• Keywords: Low vision, eye tracking, assistive technology, reading behavior, magnified screen, visual challenges, gaze patterns, line switching, reading efficiency

Research Background and Issues

• Background: Reading is essential for daily life, but it can be challenging for people with low vision. Low vision refers to a visual impairment that cannot be corrected with glasses or other standard methods. People with low vision generally read at a speed approximately three times slower than those with normal vision. Although various assistive tools are available on the market (e.g., screen magnifiers, enlarged fonts, and high-contrast settings), they are still insufficient to significantly improve the reading experience.
• Issues and Challenges:
  1. Low vision users often experience reduced visual span and difficulties with line switching when using screen magnifiers.
  2. Traditional eye tracking devices are not adequately designed for low vision users, leading to significant reductions in calibration and data collection accuracy during use.
  3. The specific gaze behaviors and visual challenges of low vision users remain under-researched.
• Research Significance: Eye tracking technology can provide precise visual behavior analysis for low vision users and has the potential to pave the way for developing more targeted assistive tools.

Proposed Solutions

• Methods and Overview:
  1. Improved Eye Tracking Calibration and Data Collection Interface: Provide adjustable calibration targets and develop a dominant-eye-based data collection strategy.
  2. Study Design: Utilize the commercial Tobii Pro Fusion eye tracker and recruit 20 low vision users and 20 normal vision users to conduct reading experiments (with and without magnification modes).
  3. Research Questions:
     • RQ1: Can commercial eye trackers reliably collect eye movement data from low vision users?
     • RQ2: How do gaze behaviors of low vision users differ from those of normal vision users?
     • RQ3: What impact do different visual conditions (visual acuity, visual field) have on the gaze behaviors of low vision users?
     • RQ4: How do different screen magnification modes affect the reading behavior of low vision users?
• Innovations: The proposed flexible calibration design and dominant-eye data collection strategy effectively improve data quality for low vision users. Eye tracking data reveal fine-grained reading behaviors of low vision users.

Research Findings

1. Validation of Eye Tracker Data Quality (RQ1):

   • After adjusting target size, the calibration accuracy of low vision users was comparable to that of normal vision users.
   • Data loss rate was slightly higher for low vision users (average 4.62% vs. 1.35%), primarily due to factors such as proximity to the screen and pupil detection failures.

2. Gaze Behavior Characteristics of Low Vision Users (RQ2):

   • Low vision users exhibited more fixation points but shorter fixation durations, indicating lower information processing efficiency.
   • Forward saccade lengths were significantly shorter than those of normal vision users, suggesting smaller visual spans and insufficient information acquisition.
   • Low vision users required more line searches during line switching, highlighting the difficulty in locating new lines.

3. Impact of Visual Conditions (RQ3):

   • Users with lower visual acuity or restricted visual fields demonstrated shorter fixation durations and smaller saccade spans.
   • Users with both low acuity and restricted visual fields showed significantly reduced information processing capabilities.

4. Impact of Different Screen Magnification Modes (RQ4):

   • Using magnifiers (whether lens-type or full-screen magnification) significantly increased the number of regressions and line searches for low vision users.
   • Larger magnification window widths improved visual span and enhanced reading speed.
   • Some users suggested that the window height should dynamically adjust to balance “reducing distractions” and “providing context.”

5. Design Implications:

   • Develop real-time visual enhancement tools (e.g., line highlighting, dynamic line spacing adjustment).
   • Improve magnifier tools to allow dynamic resizing of the magnification window based on gaze movements or implement gaze-controlled hands-free functionality.
   • Address word recognition difficulties for low vision users by introducing automatic prompts or text-to-speech features.
   • Consider personalized designs to provide user-specific visual adaptations (e.g., calibration target size, color, screen layout).

6. Experimental Limitations and Future Directions:

   • Sample diversity resulted in significant performance differences among low vision groups; future research should focus on specific subgroups.
   • Age differences between low vision and normal vision groups may introduce bias.
   • The current experiment primarily focused on reading aloud; future studies should expand to silent reading scenarios.

Conclusion

This study demonstrates the potential application of commercial eye trackers for low vision populations and reveals unique challenges through fine-grained gaze pattern analysis. The findings provide valuable practical guidance for designing assistive tools such as real-time display aids, dynamic magnifiers, and multimodal reading technologies for low vision users, while paving the way for future research in related fields.