Automatic calibration of a gaze direction algorithm from user behavior

• US 8,982,046 B2
• Filed: 06/22/2011
• Issued: 03/17/2015
• Est. Priority Date: 12/22/2008
• Status: Active Grant

• Current Assignee: Seeing Machines LTD
• Original Assignee: Seeing Machines LTD
• Inventors: Edwards, Timothy James Henry, Langdale-Smith, Nicholas John
• Primary Examiner(s): SHERMAN, STEPHEN G
• Application Number:
  US13/166,355
• Publication Number:
  US 20110310006A1
• Time in Patent Office: 1,364 Days
• Field of Search: 345/156
• US Class Current: 345/156
• CPC Class Codes: A61B 3/113 for determining or recording eye movement
• CPC Class Codes: G06F 3/013 Eye tracking input arrangements G06F3/015 takes precedence
• CPC Class Codes: G06V 40/19 Sensors therefor

Abstract

A method of calibrating the eye gaze direction of a user, the method including the steps of: (a) monitoring a user'"'"'s eye gaze direction whilst carrying out a series of predetermined tasks, each of the tasks having an expected subject gaze direction; and (b) correlating the user'"'"'s eye gaze direction with the expected direction for a statistically significant period of time; (c) calculating, from the correlating step, a series of likely eye gaze direction usage parameters associated with the user.

Claims

1. A method of calibrating the eye gaze direction of a user in a three dimensional environment, the method comprising the steps of:

(a) tracking the user'"'"'s uncalibrated eye gaze direction without the direct cooperation of the user;

(b) registering eye gaze direction measurements during a period of time when the user is likely to be looking in a direction corresponding to a known gaze direction based on the user'"'"'s probable behaviour within the tracking environment;

(c) performing a statistical analysis on the eye gaze direction measurements;

(d) correlating the user'"'"'s eye gaze direction with the known gaze directions for a statistically significant period of time; and

(e) calculating, from said correlating step, a series of likely eye gaze direction usage parameters associated with said user and applying these parameters to the eye gaze measurements to calibrate the eye gaze direction.

2. A method as claimed in claim 1 wherein, during the period of time, the user is performing a task including at least one of interacting with a computer graphical user interface or driving a vehicle.

3. A method as claimed in claim 1 further comprising the steps of:

for new users, storing first eye gaze direction usage parameters associated with a user'"'"'s first utilisation of the method;

for predetermined users of the method, determining if the user has a previous history of usage of the method; and

for users utilising the method having a previous history of usage of the method, utilising the corresponding first eye gaze direction usage parameters as an initial estimate of the calibration.

4. A method as claimed in claim 1 wherein said correlating step includes rejection of outlying data sets.

5. A method as claimed in claim 1 wherein, during the period of time, the user performs tasks including clicking on an icon in a graphical user interface for a computer.

6. A method as claimed in claim 1 wherein, during the period of time, the user performs tasks including typing within a program for a computer.

7. A method as claimed in claim 1 wherein said performing a statistical analysis includes the steps of:

(i) forming a histogram of measurements of user'"'"'s gaze direction;

(ii) determining likely peaks in the histogram measurements; and

(iii) correlating the peaks with known objects in the environment around the user.

8. A method according to claim 1 wherein items of interest include vehicle mirrors, vehicle instruments and objects viewed outside the vehicle.

9. A system for calibrating the eye gaze direction of a user in a three dimensional environment, the system including:

an eye gaze direction measuring unit for tracking the user'"'"'s uncalibrated eye gaze direction without the direct cooperation of the user; and

a processor configured to;

register eye gaze direction measurements during a period of time when the user is likely to be looking in a direction corresponding to a known gaze direction based on the user'"'"'s probable behaviour within the tracking environment;

perform a statistical analysis on the eye gaze direction measurements;

correlate the user'"'"'s eye gaze direction with the known gaze directions for a statistically significant period of time;

calculate, from the correlation, a series of likely eye gaze direction usage parameters associated with said user; and

apply these parameters to the eye gaze measurements to calibrate the eye gaze direction.

10. A method of calibrating the eye gaze direction of a user, the method comprising the steps of:

(a) measuring, without the direct cooperation of the user, the user'"'"'s uncalibrated eye gaze direction repeatedly during a time period when the user is likely to be looking in a direction corresponding to a known gaze direction based on the user'"'"'s probable behaviour within the tracking environment;

(b) building a histogram of user gaze direction values during the time period;

(c) correlating peaks in the histogram with items of interest in the environment of the tasks;

(d) calculating, from said correlating step, a series of likely eye gaze direction usage parameters associated with said user and applying these parameters to the eye gaze measurements to calibrate the eye gaze direction.

11. A method as claimed in claim 10 wherein said step (b) includes building a histogram of user gaze time verses gaze direction.

12. A method as claimed in claim 10 wherein, during the time period, the user is driving a vehicle.

13. A method as claimed in claim 10 wherein, during the time period, the user is interacting with a computer graphical display.