Optimizing Chinese character displays improves recognition and reading performance of simulated irregular phosphene...

Area of Science:

• Visual prosthetics
• Human-computer interaction
• Artificial vision

Background:

• Visual prosthetics aim to restore vision but often produce irregular phosphene maps.
• Irregular phosphene maps pose challenges for accurate visual perception and reading.

Purpose of the Study:

• To optimize Chinese character (CC) display methods for irregular phosphene maps.
• To enhance recognition and reading performance in simulated visual prosthesis conditions.

Main Methods:

• Two experiments were conducted with 20 participants (20-30 years old).
• Experiment 1 compared a projection method with a nearest neighbor search (NNS) method for CC recognition accuracy under varying irregularity levels.
• Experiment 2 assessed reading accuracy (RA) and reading efficiency (RE) using regular array correspondence and NNS methods.

Main Results:

• The NNS method significantly outperformed the projection method in CC recognition accuracy, achieving up to 81.3 ± 2.7% and 59.1 ± 5.2% for different irregularity levels.
• Reading accuracy and efficiency decreased with increased distortion levels.
• NNS significantly improved RA (from ~40% to >80%) and RE (from ~13 to >40 characters/min) for irregular text.

Conclusions:

• Optimizing display methods is crucial for improving performance with irregular phosphene arrays in visual prosthetics.
• The nearest neighbor search method demonstrates significant potential for enhancing visual perception and reading capabilities.