Correcting systematic calibration errors in eye tracking data

Patent metadata

Field	Value
Publication number	US-9782069-B2
Application number	US-201414534348-A
Country	US
Kind code	B2
Filing date	Nov 6, 2014
Priority date	Nov 6, 2014
Publication date	Oct 10, 2017
Grant date	Oct 10, 2017

Abstract

Systems and methods are provided for post-hoc correction of calibration errors in eye tracking data, which take into consideration calibration errors that result from changes in user position during a user session in which the user's fixations on a display screen are captured and recorded by an eye tracking system, and which take into consideration errors that occur when the user looks away from a displayed target item before selecting the target item.

First claim

What is claimed is:

1. A method for correcting systematic error in calibration data of an eye tracking system, comprising:
accessing recorded information that is associated with a given session in which a user performs a computer task while selecting known target items that are displayed on a display screen during the session and while using an eye tracking system to record locations of user fixations on the display screen as the user views the display screen during the session, wherein the recorded information comprises (i) fixation data that include start time, display screen location, and duration information for each of a plurality of fixations recorded during the session, (ii) selection event data that include start time and display screen location information for each of a plurality of user selection events recorded during the session; and (iii) known target location data that include display screen location information for each of a plurality of known target items that are displayed during the session;
processing the obtained information to identify a known target item selected by the user;
processing the obtained information to identify a recorded fixation, if any, which most likely corresponds to the identified known target item selected by the user, wherein a recorded fixation which was not active at a time that the identified known target item was selected by the user, is considered during said processing as a candidate fixation;
computing a distance vector between the identified recorded fixation and the corresponding identified known target item; and
storing, in a data store, the computed distance vector as calibration error information for the identified recorded fixation and the corresponding identified known target item,
wherein the accessing, processing, computing and storing are automated steps implemented by a computing device executing program code.

2. The method of claim 1 , wherein processing the obtained information to identify a recorded fixation, if any, which most likely corresponds to the identified known target item selected by the user comprises determining a recorded fixation (i) which falls within a predefined time window prior to the time that the identified known target item was selected by the user, and (ii) which is nearest to a center region of the identified known target item selected by the user.

3. The method of claim 1 , wherein processing the obtained information to identify a recorded fixation, if any, which most likely corresponds to the identified known target item selected by the user comprises determining a recorded fixation (i) which falls within a predefined time window prior to the time of selection of the identified known target item, (ii) which has a start time that is nearest in time to the time of selection of the identified known target item, but which start time is prior to the time of selection of the identified known target item by a predetermined amount of time, and (iii) which has a duration that meets a minimum predefined duration value.

4. The method of claim 1 , further comprising:
processing the obtained information to identify a recorded fixation that does not correspond to a known target item selected by the user;
computing a calibration error vector for the identified fixation by taking a weighted average of distance measures of a set of target fixations that are nearest in both space and time to the identified fixation; and
storing, in the data store, the computed calibration error vector as calibration error information for the identified fixation.

5. The method of claim 4 , further comprising using the stored calibration error information to correct the fixation locations of the recorded fixations.

6. The method of claim 4 , wherein computing a calibration error vector for the identified fixation, comprises:
processing the obtained information to determine a candidate set of user selections of known target items which occurred with a predefined time window before and after a start time of the identified fixation;
setting a fixation location of the identified fixation as an origin of a coordinate space surrounding the identified fixation;
for each quadrant of the coordinate space surrounding the identified fixation, determining from the candidate set of user selections of known target items a closest known target to the identified fixation;
for each determined known target item in each quadrant, computing an average distance vector between the determined known target item and all recorded fixations on the determined known target item;
applying a weight to each of the computed average distance vectors, wherein the average distance vectors are weighted according to the distance between the identified fixation and the selected known targets associated with the distance vectors; and
computing weighted mean of the weighted average distance vectors.

7. The method of claim 6 , further comprising excluding a recorded fixation that is deemed an outlier from average distance vector computation step.

8. An article of manufacture comprising a computer readable storage medium having program code embodied therewith, wherein the program code is executable by a computer to cause the computer to perform an automated process to correct systematic error in calibration data of an eye tracking system, wherein the automated process comprises:
accessing recorded information that is associated with a given session in which a user performs a computer task while selecting known target items that are displayed on a display screen during the session and while using an eye tracking system to record locations of user fixations on the display screen as the user views the display screen during the session, wherein the recorded information comprises (i) fixation data that include start time, display screen location, and duration information for each of a plurality of fixations recorded during the session, (ii) selection event data that include start time and display screen location information for each of a plurality of user selection events recorded during the session; and (iii) known target location data that include display screen location information for each of a plurality of known target items that are displayed during the session;
processing the obtained information to identify a known target item selected by the user;
processing the obtained information to identify a recorded fixation, if any, which most likely corresponds to the identified known target item selected by the user, wherein a recorded fixation which was not active at a time that the identified known target item was selected by the user, is considered during said processing as a candidate fixation;
computing a distance vector between the identified recorded fixation and the corresponding identified known target item; and
storing, in a data store, the computed distance vector as calibration error information for the identified recorded fixation and the corresponding identified known target item,
wherein the accessing, processing, computing and storing are automated steps implemented by a computing device executing program code.

9. The article of manufacture of claim 8 , wherein processing the obtained information to identify a recorded fixation, if any, which most likely corresponds to the identified known target item selected by the user comprises determining a recorded fixation (i) which falls within a predefined time window prior to the time that the identified known target item was selected by the user, and (ii) which is nearest to a center region of the identified known target item selected by the user.

10. The article of manufacture of claim 8 , wherein processing the obtained information to ide

Assignees

• IBM

Inventors

• Bellamy Rachel K E
• John Bonnie E
• Richards John T
• Swart Calvin B
• Thomas Jr John C
• Trewin Sharon M

Classifications

• A61B3/032

  Devices for presenting test symbols or characters, e.g. test chart projectors (A61B3/036 takes precedence) · CPC title

• A61B3/005

  Constructional features of the display · CPC title

• A61B3/0025

  characterised by electronic signal processing, e.g. eye models · CPC title

• A61B3/0091

  Fixation targets for viewing direction · CPC title

• A61B5/168

  Evaluating attention deficit, hyperactivity · CPC title