T-Space at The University of Toronto Libraries >
School of Graduate Studies - Theses >
Please use this identifier to cite or link to this item:
|Title: ||A Calibration Free Estimation of the Point of Gaze and Objective Measurement of Ocular Alignment in Adults and Infants|
|Authors: ||Model, Dmitri|
|Advisor: ||Eizenman, Moshe|
|Department: ||Electrical and Computer Engineering|
|Issue Date: ||10-Jan-2012|
|Abstract: ||Two novel personal calibration procedures that do not require active user participation are presented. These procedures, in conjunction with a state-of-the-art remote eye-gaze tracking (REGT) technology, allow estimation of the angle between the optical and visual axes (angle kappa) automatically without explicit/active user involvement. The first algorithm for the binocular estimation of angle kappa (BEAK) is based on the assumption that at each time instant both eyes look at the same point on a surface with a known geometry (e.g., a computer monitor). The sensitivity of the BEAK procedure to the geometry of the observation surface and to the noise in the estimates of the optical axis is studied both analytically and in computer simulations. Experimental results with 4 adult subjects suggest that with the current REGT technology angle kappa can be estimated with an RMS error of 0.5°.
The second personal calibration algorithm (‘calibrate and validate’, CaVa) adopts a probabilistic approach to the estimation of angle kappa in infants. Even though the presentation of visual stimuli at known positions is part of the procedure, the CaVa algorithm does not require/assume continuous fixation on the presented targets. If an infant attends to roughly half of the presented targets, angle kappa can be estimated accurately and with high confidence. In experiments with five babies, the average difference between repeated measurements of angle kappa was 0.04 ± 0.31°.
The second part of the thesis describes two methods for automated measurement of eye misalignment in adults and infants. These methods are based on the user-calibration-free (UCF) technology presented in the first part of the thesis. The first method is based on the clinical Hirschberg test. It is shown that the UCF-REGT technology can improve significantly the accuracy of the Hirschberg test by enabling the estimation of subject-specific parameters (the Hirschberg ratio and angle kappa) in infants. The maximum error in the estimation of the horizontal and vertical components of eye misalignment in five orthotropic infants was shown to be less than 1°, which is significantly better than the accuracy of a standard clinical Hirschberg test. Finally, a novel Eye-Tracker Based Test (ETBT) for the estimation of the maximum (manifest + latent) angle of deviation is presented. ETBT is based on the UCF-REGT system. ETBT allows free head movements and does not require continuous fixation on specific targets. Experiments with 22 adult subjects demonstrated a good agreement of 0.7 ± 1.7° between ETBT and the gold-standard clinical procedure—the altenate prism and cover test. A pilot study with 5 orthotropic infants and one infant with strabismus demonstrated that the ETBT can be used in infants.|
|Appears in Collections:||Doctoral|
This item is licensed under a Creative Commons License
Items in T-Space are protected by copyright, with all rights reserved, unless otherwise indicated.