Hybrid FPGA-CPU pupil tracker.

Area of Science:

• Ophthalmology
• Biomedical Engineering
• Computer Vision

Background:

• Accurate eye tracking is essential for ophthalmic applications like eye movement stabilization.
• Existing pupil tracking systems may face limitations in precision, latency, or integration capabilities.

Purpose of the Study:

• To describe and demonstrate an off-axis monocular pupil tracker for integration into ophthalmoscopes.
• To achieve sub-pixel pupil center accuracy and low processing latency for real-time eye movement stabilization.

Main Methods:

• Utilized a system comprising light-emitting diodes, a camera, a field-programmable gate array (FPGA), and a central processing unit (CPU).
• Implemented image processing algorithms including background subtraction, filtering, thresholding, and edge detection on the FPGA.
• Performed least-squares ellipse fitting to pupil edge coordinates on the CPU.

Main Results:

• Achieved sub-pixel pupil center accuracy with images having a minimum of ~32 gray levels.
• Demonstrated 0.5-1.5 μm pupil center estimation precision with 0.6-2.1 ms latency across various camera frame rates.
• Showed that tracker operation requires adjustment of only a single parameter (image intensity threshold).

Conclusions:

• The developed pupil tracker meets requirements for high precision and low latency in ophthalmic applications.
• The system's performance is primarily limited by camera download time and sensitivity.
• This technology holds promise for enhancing eye movement stabilization in ophthalmoscopes.