August 2016
Volume 16, Issue 12
Open Access
Vision Sciences Society Annual Meeting Abstract  |   September 2016
Hybrid Calibration for Eye Tracking: Smooth Pursuit Trajectory with Anchor Points
Author Affiliations
  • Quan Wang
    Yale Child Study Center, School of Medicine, Yale University
  • Erin Barney
    Yale Child Study Center, School of Medicine, Yale University
  • Carla Wall
    Yale Child Study Center, School of Medicine, Yale University
  • Lauren DiNicola
    Yale Child Study Center, School of Medicine, Yale University
  • Claire Foster
    Yale Child Study Center, School of Medicine, Yale University
  • Yeojin Ahn
    Yale Child Study Center, School of Medicine, Yale University
  • Beibin Li
    Yale Child Study Center, School of Medicine, Yale University
  • Chawarska Katarzyna
    Yale Child Study Center, School of Medicine, Yale University
  • Frederick Shic
    Yale Child Study Center, School of Medicine, Yale University
Journal of Vision September 2016, Vol.16, 1355. doi:https://doi.org/10.1167/16.12.1355
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Quan Wang, Erin Barney, Carla Wall, Lauren DiNicola, Claire Foster, Yeojin Ahn, Beibin Li, Chawarska Katarzyna, Frederick Shic; Hybrid Calibration for Eye Tracking: Smooth Pursuit Trajectory with Anchor Points. Journal of Vision 2016;16(12):1355. https://doi.org/10.1167/16.12.1355.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

Conventional eye-tracking calibration use sparse points that require saccades to all locations or pursuit trajectories that require completeness. In this work, we constructed a hybrid calibration system combining smooth trajectory and sparse points and applied it with typically developing (TD) toddlers, toddlers with Autism Spectrum Disorder (ASD) and typical adults. Employing an Eyelink eye tracker in remote mode with 500 Hz, we recorded raw pupil and corneal reflection positions. Single attempt calibrations involved a central fixation point, followed by a circular pursuit trajectory with constant speed and then four points near the screen's corners. First, we used a large circle (diameter: 20 degrees, 5.3 deg/sec) with 10 adult participants, 4 TD toddlers and 4 toddlers with ASD. Each participant had up to 30 internal calibrations. Second, we used a small circle (10 degrees, 2.6 deg/sec) with 9 toddlers with ASD, each with up to 24 internal calibrations. We validated current calibration on next internal calibration dataset. We used a weighted random sample consensus (RANSAC) method to find the best fit of least squares error mapping between the displayed stimuli and recorded raw data on eye position. We weighted the number of samples in the five points and the pursuit trajectory equally. Our results showed that three points plus the circular trajectory was comparable to five-point-calibration. With five points and pursuit trajectory together, the calibration and validation were significantly better than 5-point-calibration alone in all three participant groups and settings (p< 0.001 for calibration errors, and p< 0.05 for validation errors). With the hybrid calibration system, we created distinct eye movement events and increased sample size, which guaranteed the stability of general geometry on the screen and increased the reliability and accuracy of the screen center area. During single attempt calibration attempts, the hybrid calibration improved calibration results in toddlers and adults.

Meeting abstract presented at VSS 2016

×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×