What Is Gaze Tracking?

Gaze tracking (also called eye tracking) is the measurement of the point of visual gaze: where a person is looking at any given moment. By recording the position, direction, and movement of the eyes over time, gaze tracking systems provide insight into visual attention, cognitive load, reading behavior, and user intent. The technology combines optics, image processing, signal processing, and machine learning, and its applications range from basic vision science research to consumer products, medical diagnostics, and accessibility tools for people with severe motor impairments.

Eye Anatomy and Measurement Principles

The eye's observable features that gaze tracking systems exploit include the pupil (a dark circular aperture), the corneal reflection (a bright specular reflection of an illumination source from the curved corneal surface), and the iris limbus (the boundary between the iris and the white sclera). By tracking the relative positions of the pupil center and one or more corneal reflections (Purkinje reflections), a system can robustly estimate gaze direction as the user's head moves.

Most commercial video-based systems use near-infrared (NIR) illumination to create stable, high-contrast corneal reflections that are invisible to the user. A camera captures the eye image, and image processing algorithms locate the pupil and Purkinje reflections to compute the gaze vector, which is then mapped to a screen coordinate through a calibration procedure in which the user looks at a series of known points.

Electrooculography

Electrooculography (EOG) measures eye movements by detecting the electrical potential difference between the front (cornea, positively polarized) and back (retina, negatively polarized) of the eyeball. Electrodes placed around the eye pick up changes in this corneoretinal potential as the eye rotates. EOG does not require a camera and can operate in complete darkness or with closed eyes, making it useful for sleep studies (to distinguish REM from non-REM sleep) and for fatigue monitoring in pilots and vehicle operators. However, its spatial accuracy is lower than video-based methods, and it is susceptible to electrode drift. EOG signal processing methods are reviewed in the IEEE Transactions on Biomedical Engineering.

Saccade Detection and Signal Processing

Eye movements fall into several categories: saccades are rapid, ballistic jumps between fixation points; smooth pursuits track moving objects; fixations are relatively stable periods between saccades; and microsaccades are tiny involuntary movements during attempted fixation. Distinguishing these movement types from raw gaze coordinate time series requires signal processing algorithms that handle noise, blinks, and artifacts.

Velocity-based algorithms identify saccades when the instantaneous gaze velocity exceeds a threshold, typically 30 to 100 degrees per second. Dispersion-based algorithms identify fixations as clusters of gaze points within a small spatial window over a minimum duration. Machine learning approaches train classifiers on labeled gaze data to recognize movement types in diverse recording conditions. Research on robust classification methods is published in ACM CHI and IEEE ISMAR conference proceedings.

Virtual Reality and Augmented Reality Integration

Gaze tracking integrated into VR and AR headsets enables foveated rendering: the display renders the region at the user's gaze point at full resolution and reduces rendering quality in the periphery where the eye cannot resolve fine detail anyway. This can reduce the GPU load of VR rendering by 50 to 90 percent without a perceptible loss of image quality. Gaze data also enables natural interaction with virtual interfaces (look to select), social presence features (avatars with realistic eye contact), and cognitive workload monitoring for training applications. The IEEE VR conference covers gaze-enabled interaction and foveated rendering research extensively.

Applications

Gaze tracking serves a broad and growing set of scientific, medical, and commercial applications:

• Assistive technology: Eye-controlled communication devices allow people with ALS, locked-in syndrome, or severe cerebral palsy to type, control computers, and operate powered wheelchairs.
• Usability and UX research: Heat maps of where users look on websites, app screens, or product packaging guide interface design and advertising placement.
• Clinical diagnostics: Abnormal saccade patterns are biomarkers for neurological conditions including Parkinson's disease, schizophrenia, and concussion.
• Driver monitoring: In-cabin eye trackers detect drowsiness or distraction and trigger alerts or autonomous safety interventions.
• Foveated rendering in VR/AR: Real-time gaze data drives adaptive rendering to reduce computational demands in immersive displays.
• Reading and education research: Fixation and regression analysis reveals how skilled and struggling readers process text at the word and letter level.