Related Experiment Videos
Robust and real-time torsional eye position calculation using a template-matching technique
Danjie Zhu1, Steven T Moore, Theodore Raphan
1Department of Computer and Information Science, Brooklyn College of the City University of New York, 2900 Bedford Avenue, Brooklyn, NY 11210, USA.
Computer Methods and Programs in Biomedicine
|May 12, 2004
Summary
A new template-matching algorithm accurately calculates torsional eye position faster and with less noise than traditional methods. This advancement improves computational efficiency for eye-tracking applications.
Area of Science:
- Ophthalmology
- Computer Vision
- Biomedical Engineering
Background:
- Traditional methods for calculating eye torsion use cross-correlation of iris signatures.
- These methods often involve complex computations and can be prone to noise.
Purpose of the Study:
- To develop a novel, faster, and more accurate algorithm for computing torsional eye position.
- To reduce computational load and improve the signal-to-noise ratio in eye-tracking.
Main Methods:
- A template-matching technique was developed using iris signatures from two annuli centered on the pupil.
- The algorithm leverages the constraint of limited eye rotation (<2 degrees) between successive video frames.
- A small window of the previous reference signature is used to determine relative torsional displacement.
Main Results:
- The new algorithm significantly reduces the number of pixels required for torsion computation.
- It achieves higher accuracy and exhibits considerably less noise compared to the cross-correlation technique.
- The system can process three-dimensional eye position at 120 frames/s.
Conclusions:
- The developed template-matching algorithm offers a superior method for calculating torsional eye position.
- This advancement provides a faster, more accurate, and less noisy solution for eye-tracking systems.
- The algorithm has the potential to enhance various applications requiring precise eye movement analysis.