Optics: eye examining – vision testing and correcting – Eye examining or testing instrument – Objective type
Reexamination Certificate
2001-12-28
2003-12-09
Manuel, George (Department: 3737)
Optics: eye examining, vision testing and correcting
Eye examining or testing instrument
Objective type
Reexamination Certificate
active
06659611
ABSTRACT:
FIELD OF THE INVENTION
This invention relates to eye gaze tracking by analysis of images taken of a user's eye. The invention relates more specifically to eye gaze tracking without calibrated cameras, direct measurements of specific users' eye geometries, or requiring the user to visually track a cursor traversing a known trajectory.
BACKGROUND OF THE INVENTION
Eye gaze tracking technology has proven to be useful in many different fields, including human-computer interfaces for assisting disabled people interact with a computer. The eye gaze tracker can be used as an input device, instead of or in addition to a mouse for a personal computer, for example, helping disabled people to move a cursor on a display screen to control their environment and communicate messages. Gaze tracking can also be used for industrial control, aviation, and emergency room situations where both hands are needed for tasks other than operation of a computer but where an available computer is useful. There is also significant research interest in eye gaze tracking for babies and animals to better understand such subjects' behavior and visual processes.
There are many different schemes for detecting both the gaze direction and the point of regard, and many vendors of eye gaze tracking equipment (see for example web site http://ibs.derby.ac.uk/emed). Any particular eye gaze tracking technology should be relatively inexpensive, reliable, unobtrusive, easily learned and used and generally operator-friendly to be widely accepted. However, commercially available systems are expensive (over $10,000), complicated to install, and require a trained operator and a calibration process before each use session.
Corneal reflection eye gaze tracking systems project light toward the eye and monitor the angular difference between pupil position and the reflection of the light beam from the cornea surface. Near-infrared light is often employed, as users cannot see this light and are therefore not distracted by it. The light reflected from the eye has two major components. The first component is a ‘glint’, which is a very small and very bright virtual image of the light source reflected from the front surface of the corneal bulge of the eye; the glint is also known as the first Purkinje image. The second component is light that has entered the eye and has been reflected back out from the retina. This light serves to illuminate the pupil of the eye from behind, causing the pupil to appear as a bright disk against a darker background. This retroreflection, or “bright eye” effect familiar to flash photographers, provides a very high contrast image. An eye gaze tracking system determines the center of the pupil and the glint, and the change in the distance and direction between the two as the eye is rotated. The orientation of the eyeball can be inferred from the differential motion of the pupil center relative to the glint. The eye is often modeled as a sphere of about 12.3 mm radius having a spherical corneal bulge of about 7.4 mm radius (see “Schematic Eye” by Gullstrand, in
Visual Optics
, H. H. Emsley editor, 3
rd
ed., p. 348, Butterworth, Scarborough, Ont., 1955, which is hereby incorporated by reference). The eyes of different users will have variations from these typical values, but individual dimensional values do not generally vary significantly in the short term, and thus can be stored and used for a long period.
As shown in prior art
FIG. 1
, the main components of a corneal reflection eye gaze tracking system include a video camera sensitive to near-infrared light, a near-infrared light source (often a light-emitting diode) typically mounted to shine along the optical axis of the camera, and a computer system for analyzing images captured by the camera. The on-axis light source is positioned at or near the focal center of the camera. Image processing techniques such as intensity thresholding and edge detection identify the glint and the pupil from the image captured by the camera using on-axis light, and locate the pupil center in the camera's field of view as shown in prior art FIG.
2
.
Human eyes do not have uniform resolution over the entire field of view, nor is the portion of the retina providing the most distinct vision located precisely on the optical axis. The eye directs its gaze with great accuracy because the photoreceptors of the human retina are not uniformly distributed but instead show a pronounced density peak in a small region known as the fovea centralis. In this region, which subtends a visual angle of about one degree, the receptor density increases to about ten times the average density. The nervous system thus attempts to keep the image of the region of current interest centered accurately on the fovea as this gives the highest visual acuity. A distinction is made between the optical axis of the user's eye versus the foveal axis along which the most acute vision is achieved. As shown in prior art
FIG. 3
, the optical axis is a line going from the center of the spherical corneal bulge through the center of the pupil. The optical axis and foveal axis are offset in each eye by an inward horizontal angle of about five degrees, with a variation of about one and one half degrees in the population. The offsets of the foveal axes with respect to the optical axes of a user's eyes enable better stereoscopic vision of nearby objects. The offsets vary from one individual to the next, but individual offsets do not vary significantly in the short term. For this application, the gaze vector is defined as the optical axis of the eye. The gaze position or point of regard is defined as the intersection point of the gaze vector with the object being viewed (e.g. a point on a display screen some distance from the eye). Adjustments for the foveal axis offsets are typically made after determination of the gaze vector; a default offset angle value may be used unless values from a one-time measurement of a particular user's offset angles are available.
Unfortunately, calibration is required for all existing eye gaze tracking systems to establish the parameters describing the mapping of camera image coordinates to display screen coordinates. Different calibration and gaze direction calculation methods may be categorized by the actual physical measures they require. Some systems use physically-based explicit models that take into account eyeball radius, radius of curvature of the cornea, offset angle between the optical axis and the foveal axis, head and eye position in space, and distance between the center of the eyeball and the center of the pupil as measured for a particular user. Cameras may need to be calibrated as well, so that their precise positions and optical properties are known. Details of camera calibration are described in “A Flexible New Technique for Camera Calibration”, Z. Zhang, IEEE Transactions on Pattern Analysis and Machine Intelligence, 22(11):1330-1334, 2000, (also available as Technical Report MSR-TR-98-71 at http://research.microsoft.com/~zhang/Papers/TR98-71.pdf), hereby incorporated by reference.
During system calibration, the user may be asked to fix his or her gaze upon certain “known” points in a display. At each coordinate location, a sample of corresponding gaze vectors is computed and used to accommodate head position, screen position and size, camera position, and to adapt the system to the specific properties of the user's eye, reducing the error in the estimate of the gaze vector to an acceptable level for subsequent operation. This method is disadvantageous in that a user's flow of thought is interrupted because the gaze target has nothing to do with the work the user wishes to perform. Further, the user may also be asked to click a mouse button after visually fixating on a target, but this approach may add synchronization problems, i.e. the user could look away from the target and then click the mouse button. Also, with this approach the system would get only one mouse click for each target, so there would be no chance to avera
Amir Arnon
Flickner Myron Dale
Koons David Bruce
Morimoto Carlos Hitoshi
Manuel George
McSwain Marc D.
LandOfFree
System and method for eye gaze tracking using corneal image... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with System and method for eye gaze tracking using corneal image..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and System and method for eye gaze tracking using corneal image... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3122748