(19)
(11)EP 3 137 973 B1

(12)EUROPEAN PATENT SPECIFICATION

(45)Mention of the grant of the patent:
25.03.2020 Bulletin 2020/13

(21)Application number: 15721447.9

(22)Date of filing:  23.04.2015
(51)Int. Cl.: 
G06F 3/01  (2006.01)
G06K 9/46  (2006.01)
G06K 9/00  (2006.01)
(86)International application number:
PCT/US2015/027185
(87)International publication number:
WO 2015/167906 (05.11.2015 Gazette  2015/44)

(54)

HANDLING GLARE IN EYE TRACKING

HANDHABUNG VON BLENDUNG BEI DER AUGENVERFOLGUNG

GESTION DE L'ÉBLOUISSEMENT DANS LA POURSUITE OCULAIRE


(84)Designated Contracting States:
AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR

(30)Priority: 29.04.2014 US 201414264952

(43)Date of publication of application:
08.03.2017 Bulletin 2017/10

(73)Proprietor: Microsoft Technology Licensing, LLC
Redmond, WA 98052-6399 (US)

(72)Inventors:
  • AGRAWAL, Mudit
    Redmond, Washington 98052-6399 (US)
  • THUKRAL, Vaibhav
    Redmond, Washington 98052-6399 (US)
  • EDEN, Ibrahim
    Redmond, Washington 98052-6399 (US)
  • NISTER, David
    Redmond, Washington 98052-6399 (US)
  • SWAMINATHAN, Shivkumar
    Redmond, Washington 98052-6399 (US)

(74)Representative: Goddar, Heinz J. 
Boehmert & Boehmert Anwaltspartnerschaft mbB Pettenkoferstrasse 22
80336 München
80336 München (DE)


(56)References cited: : 
EP-A2- 1 655 687
US-A1- 2007 189 606
US-A1- 2012 229 681
US-B1- 6 714 665
US-A- 6 152 563
US-A1- 2012 105 486
US-A1- 2013 285 901
  
      
    Note: Within nine months from the publication of the mention of the grant of the European patent, any person may give notice to the European Patent Office of opposition to the European patent granted. Notice of opposition shall be filed in a written reasoned statement. It shall not be deemed to have been filed until the opposition fee has been paid. (Art. 99(1) European Patent Convention).


    Description

    BACKGROUND



    [0001] Users may interface with computing systems using a variety of input mechanisms. For example, eye gaze tracking may be utilized to interact with a graphical user interface, wherein a determined location at which a user's gaze intersects the graphical user interface may be used as a positional signal for interactions with the user interface. Gaze tracking techniques may employ one more light sources to project light onto an eye, and one or more cameras to capture images of glints of the projected light as reflected from the eye. The locations of the glints and/or the pupil in the images may be utilized to determine a pupil position indicating a gaze direction.
    US6714665 B1 describes a recognition system which obtains and analyzes images of at least one object in a scene comprising a wide field of view (WFOV) imager which is used to capture an image of the scene and to locate the object and a narrow field of view (NFOV) imager which is responsive to the location information provided by the WFOV imager and which is used to capture an image of the object.
    US2007189606 A1 describes a technique for detecting large and small non-red eye flash defects in an image is disclosed. The method comprises selecting pixels of the image which have a luminance above a threshold value and labeling neighboring selected pixels as luminous regions.
    US6152563 A discloses a system for eye-gaze direction detection that uses an infrared light emitting diode mounted coaxially with the optical axis and in front of the imaging lens of an infrared sensitive video cameras for remotely recording images of the eye of the computer operator. The disclosed system is suitable for users wearing glasses. Before pupil glint identification is performed in an image of a user's eye, the system searches for glares in the image. A region of the image is considered a glare if a certain number or more of successive pixels are found with an intensity value greater than a glare threshold. Once a glare has been found, the software will find a bounding rectangle for the glare, and then remove the glare by turning every pixel in the glare's bounding rectangle black.
    US 2012229681 A1 discloses a system and a method for automatic image glare removal. Images are acquired from at least two different locations, possibly simultaneously, and from one or more cameras, and digitized if necessary. A glare pattern is identified, typically by recognition of a fully exposed area or known test glare patterns. The images are processed to remove the glare patter by subtraction of the glare pattern from at least one image and substitution of corresponding data from a related image not identically impacted.
    US2012105486 discloses eye tracking systems and methods including such exemplary features as a display device, at least one image capture device and a processing device. The display device displays a user interface including one or more interface elements to a user. The at least one image capture device detects a user's gaze location relative to the display device. The processing device electronically analyzes the location of user elements within the user interface relative to the user's gaze location and dynamically determines whether to initiate the display of a zoom window. The disclosed gaze tracking system performs glare identification before attempting to identify pupil glints. In one example, glares are found by scanning an eye image in vertical and/or horizontal directions for pixels having a higher image intensity than some given threshold value. Groups of higher image intensity pixels are then identified and the areas of such groups are analysed to determine which groups are large enough to likely correspond to glares, e.g. caused by the presence of a user's eyeglasses.
    US2013285901 A1 discloses a system and method for tracking a gaze at a distance. A remote gaze tracking system may include an infrared lighting unit including a plurality of infrared lightings to emit an infrared light toward a user, a gaze tracking module to track a position of a face of the user, and to collect, from the tracked position of the face, an eye image including at least one reflected light among a plurality of corneal reflected lights and a lens-reflected light, the corneal reflected lights being reflected from a cornea by the emitted infrared light, and the lens-reflected light being reflected from a lens of glasses, and a processor to compare a magnitude of the lens-reflected light with a threshold in the collected eye image, and when the magnitude of the lens-reflected light is equal to or less than the threshold, to detect coordinates of a center of each of the plurality of corneal reflected lights, and to calculate a gaze position.
    EP1655687 A2 discloses a system and method for actively illuminating and monitoring a subject, such as a driver of a vehicle. The system includes a video imaging camera orientated to generate images of the subject eye(s). The system also includes first and second light sources and offset from each other and operable to illuminate the subject. The system further includes a controller for controlling illumination of the first and second light sources such that when the imaging camera, based on the brightness in a captured image exceeding a threshold brightness, detects sufficient glare in the captured image, the controller controls the first and second light sources to minimize the glare and its adverse effect on gaze tracking. This is achieved by turning off the illuminating source causing the glare.

    SUMMARY



    [0002] Embodiments are disclosed that relate to performing eye gaze tracking in the presence of sources of glare, such as eyeglasses located between an eye tracking camera and an eye being tracked. The present invention is defined by the independent claims. Preferred embodiments of the invention are set forth in the dependent claims.

    BRIEF DESCRIPTION OF THE DRAWINGS



    [0003] 

    FIG. 1 shows an embodiment of an example eye tracking environment.

    FIG. 2 is a flow diagram depicting an embodiment of a method of sequencing light sources in an eye tracking system.

    FIG. 3 is a flow diagram depicting an embodiment of a method of classifying reflections in an image from an eye tracking system.

    FIG. 4 shows an example image captured by an eye tracking system according to an embodiment of the present disclosure.

    FIG. 5 shows an example of an image captured by an eye tracking system that is processed to identify saturated regions of the image according to an embodiment of the present disclosure.

    FIGs. 6A and 6B show two views of an example light source arrangement for an eye tracking system according to an embodiment of the present disclosure.

    FIG. 7 is a block diagram of an embodiment of a computing system.

    FIG. 8 shows an example sequence of light source according to an embodiment of the present disclosure.


    DETAILED DESCRIPTION



    [0004] In an eye tracking system, camera(s) and/or light source(s) may be positioned in a location that is spaced from the eye and/or head of the user. Thus, objects may be present between the camera(s)/light source(s) and the eye, such as glasses, which may produce additional reflections of light projected by the light sources. These reflections may appear as glares in an image, and may occlude one or more of the glints and/or the pupil. Thus, such glares may interfere with eye tracking.

    [0005] As occlusion of eye tracking glints by such glares and other spurious reflections may vary with position and/or orientation of a user relative to the glint light source(s) and camera(s), different light source configurations and different types and/or thicknesses of glasses may produce different glare locations. Thus, embodiments are disclosed that relate to projecting different configurations of light sources to help identify a light source configuration that allows eye tracking to be performed without unacceptable occlusion of eye glints from glares caused by glasses and the like.

    [0006] FIG. 1 shows an example eye tracking environment 100 in which a user 102 is viewing a computing device 104 while wearing glasses 106. The computing device 104 is depicted as a tablet, but it will be understood that any other suitable computing device may utilize eye tracking. Examples include, but are not limited to, smart phones, laptops, personal computers, televisions, and wearable computing devices such as head-mounted display devices.

    [0007] Computing device 104 includes an eye tracking system comprising a plurality of light sources 108 and a camera 110. Light sources 108 may comprise, for example, a plurality of light emitting diodes (LEDs), and/or other suitable light emitting devices. In some embodiments, the light sources 108 may emit infrared light, visible light, or combinations of visible and infrared light (e.g., a subset of the light sources 108 may project infrared light and another subset of the light sources 108 may project visible light). The camera 110 may comprise any suitable imaging device, including but not limited to a depth camera, an RGB (color imaging) camera, a grayscale camera, a stereo camera pair, and/or any other suitable camera or combination of cameras. It will be understood that one or more of the light sources, the camera(s), and/or any other element of the eye tracking system may be integrated within a single computing device, housed separately from the computing device, or arranged in any combination thereof.

    [0008] As illustrated by the dashed lines in FIG. 1, each light source 108 may emit light toward an eye of the user 102. The camera 110 may capture images of the eye of the user 102 that include reflections from the eye of the light projected from the light sources 108. Based on a location of the reflections of the projected light in the image from the camera 110 compared to a pupil (or iris, or other suitable eye structure) of the user's eye, a direction of gaze may be determined. This may allow a gaze to be projected from the eye, such that a location at which the projected gaze intersects a user interface or a real-world object may be determined. This may allow a user to interact with a computing device via gaze. Further, changes in gaze location over time may be used as gesture inputs for a computing device.

    [0009] FIG. 2 shows a flow diagram depicting an embodiment of a method 200 for tracking eye movements that may help to achieve robust eye tracking performance in the presence of glasses or other such structure between the light source(s)/camera(s) and the user's eye. Method 200 may be performed by an eye tracking system in a computing device, such as computing device 104 of FIG. 1.

    [0010] At 202, method 200 includes acquiring eye tracking data. As described above, eye tracking may be performed by emitting light (e.g., infrared light) toward an eye of a user and capturing images of the light as reflected from the eye of the user. However, as light also may be reflected from eyeglasses or other transparent or semi-transparent optical structures between the light sources and the user's eye, glares may arise that occlude the reflections of the light from the user's eye.

    [0011] Thus, as indicated at 204, method 200 may include iteratively projecting light from different combinations of light sources, and at 206, capturing an image of the eye during the projection of each different combination of light sources, as indicated at 206. These processes may involve, for example, projecting light from different numbers of light sources in the different combinations and/or projecting light from light sources having different positions/orientations. As a more specific example, FIG. 8 schematically illustrates an eye tracking system that include four light sources 802a-802d, wherein illuminated light sources are shown by diagonal lines within a box representing a light source. Iteratively projecting light from different combinations of light sources may include projecting light from all light sources, (as shown at time T1); then from different combinations of three light sources (as shown at times T2, T3, T4, and T5); and then from different combinations of two light sources (as shown at times T6 and T7) or just one light source (not shown in the Figure). It is to be understood that such a cycle of light source projections may be performed in any suitable order. For example, combinations with greater numbers of light sources illuminated may be tried before those with lesser numbers of light sources where a most accurate gaze determination is desired, while those with lesser numbers may be tried before those with greater numbers where power savings is desired, or where the glass surfaces tend to produce more glares.

    [0012] Further, in some embodiments, an order of combinations of light sources to project may optionally be selected based on a head/HMD position and/or an orientation/position of the light sources, as indicated at 208. For example, it may be known that particular numbers and/or patterns of light sources may produce fewer occlusions when a head is positioned at a given angle. By selecting a next combination based on the above-described information, the different combinations of light sources may be iteratively cycled in an intelligent manner to increase the likelihood that a suitable combination of light sources may be utilized in an early iteration, thereby reducing the amount of time spent cycling through different light source combinations. In this way, the eye tracking system may estimate which combination of light sources will produce the lowest amount of occlusion and iteratively project light from the different combinations of light sources in an order that is based upon the estimation. It is to be understood that in other embodiments, the combination of light sources may be selected based upon an amount of occlusion in an image, as described below.

    [0013] At 210, method 200 includes determining whether unacceptable occlusion exists in the image for each tested light source combination, and at 212, selecting a combination of light sources for performing eye tracking. As indicated at 214, a light source combination may be selected based on an amount of occlusion detected in the image. In some embodiments, the iterative testing of each combination may cease upon identification and selection of a suitable combination, while in other embodiments a full set of combinations may be tested before selecting one. As part of the testing of each combination, for a given light source configuration, glares may either be matched to their corresponding glints, or occlusion metrics may be obtained between the glares and the pupil or glints. In the case of high occlusion (e.g., occlusion above a threshold), the next light source configuration may be chosen from the sequence. The process may then repeat until unoccluded or partially occluded pupil-glints are obtained with high confidence scores. This configuration may then be utilized across future frames until a next occlusion is detected, when the configurations are again cycled through until a suitable light source configuration is again determined.

    [0014] Method 200 further includes, at 216, projecting light via the selected combination of light sources, and at 218 tracking a gaze location of one or more eyes by detecting light from the light sources as reflected from the eye(s). Further, at 220, method 200 includes performing an action responsive to the eye tracking. The eye tracking may be used to perform any suitable action. For example, the eye tracking may be utilized to detect eye gestures, to detect position signals for a graphical user interface, etc.

    [0015] The determination of unacceptable amounts of occlusion of eye glint reflections by glares may be determined in any suitable manner. FIG. 3 shows a flow diagram depicting an example embodiment of a method 300 for classifying reflections and/or glare or other interference in images captured by a camera of an eye tracking system. It will be understood that method 300 may be performed by a computing device, such as computing device 104 of FIG. 1, configured to process images in an eye tracking system.

    [0016] At 302, method 300 includes receiving image data from a camera. The camera may be integrated in a computing device or externally/remotely positioned relative to the computing device. Method 300 further includes, at 304, detecting saturated regions in the received image. For example, the image may be analyzed to determine pixels in the image with a saturation value that is higher than a threshold.

    [0017] As glares may result from specular reflections from glasses or other smooth structures, the glares may have highly saturated cores, similar to the intensity distribution of the light source itself. As such, glares formed from the light projected from light sources used in the eye tracking system may have a pattern of high intensity at the center, which dissipates abruptly moving away from the center, sometimes resulting in the appearance of flares. From such properties, glares formed from reflections of projections from the light sources may be differentiated from reflections of light off of the user's eye(s) and from other diffused reflections caused due to the presence of other IR sources in the surroundings.

    [0018] FIG. 4 shows an example depiction of an image 400 captured by a camera of an eye tracking system, and shows a view of a user 402 wearing glasses 404. Light of the eye tracking system light sources (as well as ambient light sources) may be reflected by the glasses 404, as well as by a pupil of an eye 406 of the user 402. Such reflections from the glasses 404 may result in glares 408, while reflections from the eye may result in glints 410, illustrated as four uniformly spaced dots in a region of the pupil of eye 406. While the glints 410 appear as small, substantially circular dots, the glares 408 may have a flared, star-like shape.

    [0019] Returning to FIG. 3, method 300 may include identifying and selecting saturated pixels of the image, and performing a foreground distance transform of the saturated pixels of the image, as indicated at 306, such that an intensity of a pixel after the foreground distance transform is a function of a distance from the boundary of the reflection. This may help to provide an indication of contours of glare candidates based upon a size of a saturated region and/or the contours of the saturated region. For example, a saturated region that is larger than a threshold size may be considered to be a glare candidate, while saturated regions that are smaller than a threshold size may not be considered to be a glare candidate.

    [0020] At 308, method 300 includes removing noise in the image, for example, by removing contours with a distance value that is lower than a distance threshold. In this way, the flared contours of the glares/glare candidates may be smoothed. Further, at 310, method 300 includes determining a bounding box for each remaining saturated region (e.g., the cores of the glares/glare candidates determined at 308). The size of the bounding box may be selected to have a value that enables the box to include a percentage of thresholded saturated pixels, as indicated at 312. For example, a bounding box may be formed around a core of a glare/glare candidate and a size of the bounding box may be increased until the percentage of saturated pixels in the bounding box exceeds some threshold. This resizing may help to ensure that a box is placed around each saturated region. In case of a glare, the box includes a saturated center, while in case of false positives (e.g., non-glares), saturated pixels are spread randomly throughout the box. Turning briefly to FIG. 5, a processed version of the image 400 of FIG. 4 is shown, in which saturated regions 502 (e.g., glare candidates) are surrounded by bounding boxes 504.

    [0021] Returning to FIG. 3, method 300 includes, at 314, fitting a statistical distribution to a first saturated region. For example, a Gaussian model or other statistical distribution model may be fit to detected glare centers to form normalized distribution of saturated pixels in a region of the glare candidates. A parameter of the fit of the statistical distribution for each saturated region/glare candidate then may be compared to a threshold condition. For example, a Gaussian modeling error may be determined for the Gaussian model fit to that saturated region, and a comparison of the error to a threshold error may be determined at 316. If the parameter meets the threshold (e.g., if the modeling error is below a threshold), then it may be determined at 318 that the region is a glare, and the method may proceed to 320, where it is determined whether all saturated regions have been analyzed. For example, glare candidates 506a, 506b, 506c, 506d, and 506e in FIG. 5 may be classified as glares due to the distribution of saturated pixels within the associated boxes exhibiting glare-like features, such as the concentration of saturated pixels in the central region and flares protruding at regularly spaced peripheral regions. Where it is determined at 316 that the parameter does not meet the threshold, then the method may proceed to 320 without classifying the saturated region as a glare (e.g., glare candidates 506f, 506g, 506h, 506i, 506j, and 506k may not be classified as glares due to a lack of a saturated core and/or absence of other glare features).

    [0022] At 320, if it is determined that all saturated regions have not been analyzed (e.g., "NO" at 320), then method 300 comprises iteratively performing the processes of 316, 318 and 320 until all saturated regions have been analyzed. If all saturated regions have been analyzed (e.g., "YES" at 320), then method 300 comprises, at 324, to determine a level of occlusion based on a number and/or locations of saturated regions classified as glares. For example, a level of occlusion may be based upon a size of the glares, the number of the glares, and/or how close the glares are to a pupil of the eye/glints reflected from the pupil of the eye.

    [0023] The various thresholds described above with regard to method 300 (e.g., the distance threshold at 308, the threshold percentage at 312, and the threshold condition at 316) may be predetermined and/or selected based upon statistical data. In additional or alternative embodiments, one or more of the thresholds may be determined via a learning algorithm (e.g., utilizing a classifier). For example, determining the threshold(s) via the learning algorithm may include dynamically altering the threshold(s) over time based upon measured/recorded data for a particular user, environment, lighting arrangement, and/or other suitable condition. Upon determining the thresholds using a classifier, a number of other features (e.g., a quadratic fit error, a position relative to eye corners, dissipation gradient, etc.) may be added to optimize the separation between the glares and the non-glares in the analyzed image.

    [0024] FIGS. 6A and 6B show different views of an example light source arrangement 600 of an eye tracking system. In the front view of FIG. 6A, the individual light sources 602 are illustrated as being arranged around a housing structure 604. In some embodiments, the housing structure 604 may include, be integrated within, and/or be mounted to a camera of the eye tracking system. In other embodiments, the housing structure 604 may be configured to be mounted onto other elements. As illustrated, each light source 602 may be positioned in a different location relative to other light sources. In this way, light projected from each light source 602 may be directed to a different location and/or arrive at a particular location at a different angle than light projected from other light sources in the light source arrangement 600. This may allow different combinations of light sources to be used to form reflections from the eye to avoid occlusions from glares, as described above.

    [0025] Further, as shown in the oblique view of the light source arrangement 600 illustrated in FIG. 6B, one or more of the light sources 602 may be oriented differently from other light sources in the arrangement. The dashed arrows indicate the direction of light emitted from each of the light sources 602. In this way, light projected from each light source 602 may be directed to a different location and/or arrive at a particular location from a different angle than light projected from other light sources in the light source arrangement 600.

    [0026] Occlusion of pupil glints in eye tracking images may be based on classifying reflections on the optical structure based on their features like location, size, intensity distribution, and mapping to the light sources. By providing a light source arrangement including light sources that direct light from different locations/angles, the light sources may be iteratively turned on/off to generate different combinations of light source projections in an eye tracking system. Analyzing images captured during projection of light from each combination of light sources may identify glares (e.g., determine a location of glares relative to the eye) and/or match glares to particular light sources/light source combinations. Accordingly, a light source combination that produces unoccluded pupil glints that are obtained with a high confidence score, a fewest number of occlusions of the eye/glints reflected from the eye, and/or otherwise produces a suitable eye tracking image may be selected for performing eye tracking. Selecting a particular light source combination for a given user/environment may enable the system to operate in a broader range of conditions, including conditions in which optical structures, such as glasses, are present between the eye tracking camera/light sources and the eye being tracked.

    [0027] In some embodiments, the methods and processes described herein may be tied to a computing system of one or more computing devices. In particular, such methods and processes may be implemented as a computer-application program or service, an application-programming interface (API), a library, and/or other computer-program product.

    [0028] FIG. 7 schematically shows a non-limiting embodiment of a computing system 700 that can enact one or more of the methods and processes described above. Computing system 700 is shown in simplified form. Computing system 700 may take the form of one or more personal computers, server computers, tablet computers, home-entertainment computers, network computing devices, gaming devices, mobile computing devices, mobile communication devices (e.g., smart phone), wearable computing devices, and/or other computing devices. For example, computing system 700 may be an example of computing device 104 of FIG. 1 and/or may perform the methods described in FIGS. 2 and 3.

    [0029] Computing system 700 includes a logic device 702 and a storage device 704. Computing system 700 may optionally include a display subsystem 706, input subsystem 708, communication subsystem 710, and/or other components not shown in FIG.7.

    [0030] Logic device 702 includes one or more physical devices configured to execute instructions. For example, the logic device may be configured to execute instructions that are part of one or more applications, services, programs, routines, libraries, objects, components, data structures, or other logical constructs. Such instructions may be implemented to perform a task, implement a data type, transform the state of one or more components, achieve a technical effect, or otherwise arrive at a desired result.

    [0031] The logic device 702 may include one or more processors configured to execute software instructions. Additionally or alternatively, the logic device may include one or more hardware or firmware logic devices configured to execute hardware or firmware instructions. Processors of the logic device may be single-core or multi-core, and the instructions executed thereon may be configured for sequential, parallel, and/or distributed processing. Individual components of the logic device optionally may be distributed among two or more separate devices, which may be remotely located and/or configured for coordinated processing. Aspects of the logic device may be virtualized and executed by remotely accessible, networked computing devices configured in a cloud-computing configuration.

    [0032] Storage device 704 includes one or more physical devices configured to hold instructions executable by the logic device to implement the methods and processes described herein. When such methods and processes are implemented, the state of storage device 704 may be transformed-e.g., to hold different data.

    [0033] Storage device 704 may include removable and/or built-in devices. Storage device 704 may include optical memory (e.g., CD, DVD, HD-DVD, Blu-Ray Disc, etc.), semiconductor memory (e.g., RAM, EPROM, EEPROM, etc.), and/or magnetic memory (e.g., hard-disk drive, floppy-disk drive, tape drive, MRAM, etc.), among others. Storage device 704 may include volatile, nonvolatile, dynamic, static, read/write, read-only, random-access, sequential-access, location-addressable, file-addressable, and/or content-addressable devices.

    [0034] It will be appreciated that storage device 704 includes one or more physical devices. However, aspects of the instructions described herein alternatively may be propagated by a communication medium (e.g., an electromagnetic signal, an optical signal, etc.) that is not held by a physical device for a finite duration.

    [0035] Aspects of logic device 702 and storage device 704 may be integrated together into one or more hardware-logic components. Such hardware-logic components may include field-programmable gate arrays (FPGAs), program- and application-specific integrated circuits (PASIC / ASICs), program- and application-specific standard products (PSSP / ASSPs), system-on-a-chip (SOC), and complex programmable logic devices (CPLDs), for example.

    [0036] The terms "module," "program," and "engine" may be used to describe an aspect of computing system 700 implemented to perform a particular function. In some cases, a module, program, or engine may be instantiated via logic device 702 executing instructions held by storage device 704. It will be understood that different modules, programs, and/or engines may be instantiated from the same application, service, code block, object, library, routine, API, function, etc. Likewise, the same module, program, and/or engine may be instantiated by different applications, services, code blocks, objects, routines, APIs, functions, etc. The terms "module," "program," and "engine" may encompass individual or groups of executable files, data files, libraries, drivers, scripts, database records, etc.

    [0037] It will be appreciated that a "service", as used herein, is an application program executable across multiple user sessions. A service may be available to one or more system components, programs, and/or other services. In some implementations, a service may run on one or more server-computing devices.

    [0038] When included, display subsystem 706 may be used to present a visual representation of data held by storage device 704. This visual representation may take the form of a graphical user interface (GUI). As the herein described methods and processes change the data held by the storage device, and thus transform the state of the storage device, the state of display subsystem 706 may likewise be transformed to visually represent changes in the underlying data. Display subsystem 706 may include one or more display devices utilizing virtually any type of technology. Such display devices may be combined with logic device 702 and/or storage device 704 in a shared enclosure, or such display devices may be peripheral display devices.

    [0039] Input subsystem 708 may comprise or interface with one or more user-input devices such as an eye tracking system (e.g., the eye tracking system of computing device 104 in FIG. 1), keyboard, mouse, touch screen, handwriting pointer device, or game controller. In some embodiments, the input subsystem may comprise or interface with selected natural user input (NUI) componentry. Such componentry may be integrated or peripheral, and the transduction and/or processing of input actions may be handled on- or off-board. Example NUI componentry may include a microphone for speech and/or voice recognition; an infrared, color, stereoscopic, and/or depth camera for machine vision and/or gesture recognition; a head tracker, eye tracker, accelerometer, and/or gyroscope for motion detection and/or intent recognition; as well as electric-field sensing componentry for assessing brain activity. For example, the input subsystem may comprise an eye tracking system and/or a portion of an eye tracking system utilized to perform the methods 200 and/or 300 of FIGS. 2 and 3.

    [0040] When included, communication subsystem 710 may be configured to communicatively couple computing system 700 with one or more other computing devices. Communication subsystem 710 may include wired and/or wireless communication devices compatible with one or more different communication protocols. As non-limiting examples, the communication subsystem may be configured for communication via a wireless telephone network, or a wired or wireless local- or wide-area network. In some embodiments, the communication subsystem may allow computing system 700 to send and/or receive messages to and/or from other devices via a network such as the Internet.

    [0041] It will be understood that the configurations and/or approaches described herein are exemplary in nature, and that these specific embodiments or examples are not to be considered in a limiting sense, because numerous variations are possible. The specific routines or methods described herein may represent one or more of any number of processing strategies. As such, various acts illustrated and/or described may be performed in the sequence illustrated and/or described, in other sequences, in parallel, or omitted. Likewise, the order of the above-described processes may be changed.

    [0042] The scope of the present invention is defined by the appended claims.


    Claims

    1. An eye tracking method comprising:
    iteratively testing different combinations of light sources (108) of a plurality of light sources (108), wherein in each iteration a respective one of the different combinations of light sources (108) is tested by:

    projecting (204) light toward an eye of a user from the respective one of the different combinations of light sources (108);

    capturing (206) an image of the eye with a camera (119) during projection of the respective one of the different combinations of light sources (108);

    detecting (304) saturated regions in the image;

    determining (310) a bounding box for each core of the saturated regions;

    fitting (314) a statistical distribution to each saturated region within each bounding box;

    classifying (318) a saturated region within a bounding box as a glare if a parameter of the statistical distribution fit to the saturated region meets a threshold statistical distribution fit condition;

    determining (324) a level of occlusion arising from a transparent or semi-transparent optical structure positioned between the eye and the camera (119) based on number and/or locations of glares in the image;

    selecting (212) the respective one of the different combinations of light sources (108) as a suitable combination of light sources for performing eye tracking if the determined level of occlusion shows that the respective one of the different combinations of light sources (108) produces unoccluded or partially occluded pupil glints that are obtained with a high confidence score; and

    ceasing the iterative testing upon selection of the respective one of the different combinations of light sources (108) as the suitable combination of light sources for performing eye tracking; and

    after the iterative testing, projecting (216) light via the selected suitable combination of light sources (108) for performing the eye tracking.


     
    2. An eye tracking method comprising:

    iteratively projecting (204) light toward an eye of a user from different combinations of light sources (108) of a plurality of light sources (108);

    capturing (206) an image of the eye with a camera (119) during projection of each of the different combinations of light sources (108);

    analyzing each of the captured images by:

    detecting (304) saturated regions in the image;

    determining (310) a bounding box for each core of the saturated regions;

    fitting (314) a statistical distribution to each saturated region within each bounding box;

    classifying (318) a saturated region within a bounding box as a glare if a parameter of the statistical distribution fit to the saturated region meets a threshold statistical distribution fit condition; and

    determining (324) a level of occlusion arising from a transparent or semi-transparent optical structure positioned between the eye and the camera (119) based on number and/or locations of glares;

    after each of the captured images is analyzed, selecting (212) a suitable combination of light sources (108) for performing eye tracking from the different combinations of light sources (108) based on the level of occlusion determined for each of the captured images; and

    projecting (216) light via the selected suitable combination of light sources (108) for performing the eye tracking.


     
    3. The method of claim 1 or 2, wherein determining (310) a bounding box for each core of the saturated regions includes increasing (312) a size of the bounding box until a percentage of saturated pixels in the bounding box meets a threshold bounding box condition.
     
    4. The method of claim 3, further comprising performing (306) a foreground distance transform to the image and removing (308) contours in the image having a distance value that is less than a distance threshold.
     
    5. The method of claim 4, wherein one or more of the distance threshold, the threshold bounding box condition, and the threshold statistical distribution fit condition is determined via a learning algorithm.
     
    6. The method of claim 1, wherein iteratively testing different combinations of light sources (108) includes iteratively testing the different combinations of light sources (108) in an order that is based upon a predicted occlusion for each of one or more combinations.
     
    7. The method of claim 6, further comprising detecting a head pose, and determining (208) an estimation of which combination of light sources (108) will produce the lowest amount of occlusion based upon the head pose.
     
    8. The method of claim 1 or 2, further comprising designating a saturated point in an image as a glare and matching the glare with a selected light source (108) in a first combination of light sources (108) based on a determination of whether the glare is present in a next image captured during projection of a second, different combination of light sources (108) that does not include the selected light source (108).
     
    9. The method of claim 1 or 2, wherein one or more of the plurality of light sources (108) are oriented differently from one another, and wherein, when dependent on claim 1, iteratively testing different combinations of light sources (108) comprises iteratively testing different combinations of light sources (108) having different combinations of orientations, or wherein, when dependent on claim 2, iteratively projecting light from different combinations of light sources (108) comprises iteratively projecting light from different combinations of light sources (108) having different combinations of orientations.
     
    10. An eye tracking system comprising:

    a plurality of light sources (108);

    a camera (119) configured to capture an image of light from the light sources as reflected from an eye of a user;

    a logic device (702); and

    a storage device (704) storing instructions executable by the logic device (702);

    wherein the logic device (702), when executing the instructions of the storage device (704), is configured to perform a method of any of claims 1-9.


     


    Ansprüche

    1. Ein Augenverfolgungsverfahren, das Folgendes aufweist:
    iteratives Testen verschiedener Kombinationen von Lichtquellen (108) einer Vielzahl von Lichtquellen (108), wobei in jeder Iteration jeweils eine der verschiedenen Kombinationen von Lichtquellen (108) getestet wird durch:

    Projizieren (204) von Licht in Richtung eines Auges eines Benutzers von der jeweiligen der verschiedenen Kombinationen von Lichtquellen (108);

    Erfassen (206) eines Bildes des Auges mit einer Kamera (119) während der Projektion der jeweiligen der verschiedenen Kombinationen von Lichtquellen (108);

    Erfassen (304) gesättigter Regionen im Bild;

    Bestimmen (310) eines Begrenzungsrahmens für jeden Kern der gesättigten Regionen;

    Anpassen (314) einer statistischen Verteilung an jede gesättigte Region innerhalb jedes Begrenzungsrahmens;

    Klassifizieren (318) einer gesättigten Region innerhalb eines Begrenzungsrahmens als eine Blendung, wenn ein Parameter der statistischen Verteilung, der an die gesättigte Region angepasst ist, eine statistische Verteilungsanpassungs-Schwellenwertbedingung erfüllt;

    Bestimmen (324) eines Verdeckungsgrades, der sich aus einer transparenten oder halbtransparenten optischen Struktur ergibt, die zwischen dem Auge und der Kamera (119) positioniert ist, basierend auf der Anzahl und/oder den Positionen von Blendungen im Bild;

    Auswählen (212) der jeweiligen der verschiedenen Kombinationen von Lichtquellen (108) als eine geeignete Kombination von Lichtquellen zur Durchführung der Augenverfolgung, wenn der ermittelte Grad der Verdeckung zeigt, dass die jeweilige der verschiedenen Kombinationen von Lichtquellen (108) nicht oder teilweise verdeckte Pupillenschimmer erzeugt, die mit einem hohen Vertrauenswert erhalten werden; und

    Einstellen der iterativen Prüfung bei der Auswahl der jeweiligen der verschiedenen Kombinationen von Lichtquellen (108) als geeignete Kombination von Lichtquellen zur Durchführung der Augenverfolgung; und

    nach dem iterativen Test, Projizieren (216) von Licht über die ausgewählte geeignete Kombination von Lichtquellen (108) zur Durchführung der Augenverfolgung.


     
    2. Ein Augenverfolgungsverfahren, das Folgendes aufweist:

    iteratives Projizieren (204) von Licht auf das Auge eines Benutzers aus verschiedenen Kombinationen von Lichtquellen (108) einer Vielzahl von Lichtquellen (108); Erfassen (206) eines Bildes des Auges mit einer Kamera (119) während der Projektion jeder der verschiedenen Kombinationen von Lichtquellen (108);

    Analysieren jedes der aufgenommenen Bilder durch:

    Erkennen (304) gesättigter Regionen im Bild;

    Bestimmen (310) eines Begrenzungsrahmens für jeden Kern der gesättigten Regionen;

    Anpassen (314) einer statistischen Verteilung an jede gesättigte Region innerhalb jedes Begrenzungsrahmens;

    Klassifizieren (318) einer gesättigten Region innerhalb eines Begrenzungsrahmens als eine Blendung, wenn ein Parameter der statistischen Verteilung, der an die gesättigte Region angepasst ist, eine statistische Verteilungsanpassungs-Schwellenwertbedingung erfüllt; und

    Bestimmen (324) eines Verdeckungsgrades, der sich aus einer transparenten oder halbtransparenten optischen Struktur zwischen Auge und Kamera (119) ergibt, auf der Grundlage der Anzahl und/oder der Lage der Blendungen;

    nachdem jedes der aufgenommenen Bilder analysiert wurde, Auswählen (212) einer geeigneten Kombination von Lichtquellen (108) zur Durchführung der Augenverfolgung aus den verschiedenen Kombinationen von Lichtquellen (108) auf der Grundlage des für jedes der aufgenommenen Bilder bestimmten Verdeckungsgrades; und

    Projizieren (216) von Licht über die ausgewählte geeignete Kombination von Lichtquellen (108) zur Durchführung der Augenverfolgung.


     
    3. Verfahren nach Anspruch 1 oder 2, bei dem Bestimmen (310) eines Begrenzungsrahmens für jeden Kern der gesättigten Regionen Erhöhen (312) einer Größe des Begrenzungsrahmens aufweist, bis ein Prozentsatz der gesättigten Pixel in dem Begrenzungsrahmen eine Schwellenwert-Begrenzungsrahmen-Bedingung erfüllt.
     
    4. Verfahren nach Anspruch 3, das weiterhin Durchführen (306) einer Vordergrunddistanztransformation in das Bild und Entfernen (308) von Konturen im Bild mit einem Abstandswert, der kleiner als ein Abstands-Schwellenwert ist, aufweist.
     
    5. Verfahren nach Anspruch 4, wobei eines oder mehrere des Abstands-Schwellenwerts, der Schwellenwert-Begrenzungsrahmen-Bedingung und der statistischen Verteilungsanpassungs-Bedingung des Schwellenwertes über einen Lernalgorithmus bestimmt werden.
     
    6. Verfahren nach Anspruch 1, wobei iteratives Testen verschiedener Kombinationen von Lichtquellen (108) iteratives Testen der verschiedenen Kombinationen von Lichtquellen (108) in einer Reihenfolge aufweist, die auf einer vorhergesagten Verdeckung für jede von einer oder mehreren Kombinationen basiert.
     
    7. Verfahren nach Anspruch 6, das ferner Erfassen einer Kopfhaltung und Bestimmen (208) einer Abschätzung, welche Kombination von Lichtquellen (108) das niedrigste Maß an Verdeckung auf der Grundlage der Kopfhaltung erzeugt, aufweist.
     
    8. Verfahren nach Anspruch 1 oder 2, aufweisend ferner Bezeichnen eines gesättigten Punktes in einem Bild als Blendung und Anpassen der Blendung an eine ausgewählte Lichtquelle (108) in einer ersten Kombination von Lichtquellen (108) auf der Grundlage einer Bestimmung, ob die Blendung in einem nächsten Bild vorhanden ist, das während der Projektion einer zweiten, anderen Kombination von Lichtquellen (108) aufgenommen wird, die die ausgewählte Lichtquelle (108) nicht enthält.
     
    9. Verfahren nach Anspruch 1 oder 2, wobei eine oder mehrere der mehreren Lichtquellen (108) unterschiedlich orientiert sind und wobei, wenn es von Anspruch 1 abhängig ist, iteratives Testen unterschiedlicher Kombinationen von Lichtquellen (108) iteratives Testen unterschiedlicher Kombinationen von Lichtquellen (108) mit unterschiedlichen Kombinationen von Orientierungen aufweist, oder wobei, wenn es von Anspruch 2 abhängig ist, iteratives Projizieren von Licht von unterschiedlichen Kombinationen von Lichtquellen (108) iteratives Projizieren von Licht von unterschiedlichen Kombinationen von Lichtquellen (108) mit unterschiedlichen Kombinationen von Orientierungen aufweist.
     
    10. Ein Augenverfolgungssystem, aufweisend:

    eine Vielzahl von Lichtquellen (108);

    eine Kamera (119), die so konfiguriert ist, dass sie ein Bild von Licht von den Lichtquellen einfängt, wie es von einem Auge eines Benutzers reflektiert wird;

    eine Logikvorrichtung (702); und

    eine Speichervorrichtung (704), die von der Logikvorrichtung (702) ausführbare Befehle speichert;

    wobei die Logikvorrichtung (702) bei der Ausführung der Befehle der Speichervorrichtung (704) so konfiguriert ist, dass sie ein Verfahren nach einem der Ansprüche 1-9 ausführt.


     


    Revendications

    1. Procédé de poursuite oculaire comprenant les étapes ci-dessous consistant à :
    tester de manière itérative différentes combinaisons de sources de lumière (108) d'une pluralité de sources de lumière (108), dans lequel, à chaque itération, une combinaison respective des différentes combinaisons de sources de lumière (108) est testée en mettant en œuvre les étapes ci-dessous consistant à :

    projeter (204) de la lumière vers un œil d'un utilisateur à partir de la combinaison respective des différentes combinaisons de sources de lumière (108) ;

    capturer (206) une image de l'œil avec une caméra (119) pendant la projection de la combinaison respective des différentes combinaisons de sources de lumière (108) ;

    détecter (304) des régions saturées dans l'image ;

    déterminer (310) un rectangle englobant pour chaque noyau des régions saturées ;

    ajuster (314) une distribution statistique à chaque région saturée au sein de chaque rectangle englobant ;

    classer (318) une région saturée au sein d'un rectangle englobant comme un éblouissement, si un paramètre de l'ajustement de distribution statistique à la région saturée satisfait une condition d'ajustement de distribution statistique de seuil ;

    déterminer (324) un niveau d'occlusion résultant d'une structure optique transparente ou semi-transparente positionnée entre l'œil et la caméra (119) sur la base du nombre et/ou des emplacements d'éblouissements dans l'image ;

    sélectionner (212) la combinaison respective des différentes combinaisons de sources de lumière (108) en tant qu'une combinaison appropriée de sources de lumière pour mettre en œuvre une poursuite oculaire si le niveau d'occlusion déterminé montre que la combinaison respective des différentes combinaisons de sources de lumière (108) produit des scintillements de pupille non occlus ou partiellement occlus qui sont obtenus avec un score de confiance élevé ; et

    interrompre le test itératif suite à la sélection de la combinaison respective des différentes combinaisons de sources de lumière (108) en tant que la combinaison appropriée pour la mise en œuvre d'une poursuite oculaire ; et

    après le test itératif, projeter (216) de la lumière par l'intermédiaire de la combinaison appropriée sélectionnée de sources de lumière (108) pour mettre en œuvre la poursuite oculaire.


     
    2. Procédé de poursuite oculaire comprenant les étapes ci-dessous consistant à :

    projeter de manière itérative (204) de la lumière vers un œil d'un utilisateur à partir de différentes combinaisons de sources de lumière (108) d'une pluralité de sources de lumière (108) ;

    capturer (206) une image de l'œil avec une caméra (119) pendant la projection de chacune des différentes combinaisons de sources de lumière (108) ;

    analyser chacune des images capturées en mettant en œuvre les étapes ci-dessous consistant à :

    détecter (304) des régions saturées dans l'image ;

    déterminer (310) un rectangle englobant pour chaque noyau des régions saturées ;

    ajuster (314) une distribution statistique à chaque région saturée au sein de chaque rectangle englobant ;

    classer (318) une région saturée au sein d'un rectangle englobant comme un éblouissement, si un paramètre de l'ajustement de distribution statistique à la région saturée satisfait une condition d'ajustement de distribution statistique de seuil ;

    déterminer (324) un niveau d'occlusion résultant d'une structure optique transparente ou semi-transparente positionnée entre l'œil et la caméra (119) sur la base du nombre et/ou des emplacements d'éblouissements

    suite à l'analyse de chacune des images capturées, sélectionner (212) une combinaison appropriée de sources de lumière (108) pour mettre en œuvre la poursuite oculaire à partir des différentes combinaisons de sources de lumière (108) sur la base du niveau d'occlusion déterminé pour chacune des images capturées ; et

    projeter (216) de la lumière par l'intermédiaire de la combinaison appropriée sélectionnée de sources de lumière (108) pour mettre en œuvre la poursuite oculaire.


     
    3. Procédé selon la revendication 1 ou 2, dans laquelle l'étape de détermination (310) d'un rectangle englobant pour chaque noyau des régions saturées comprend l'étape consistant à augmenter (312) une taille du rectangle englobant jusqu'à ce qu'un pourcentage de pixels saturés dans le rectangle englobant satisfasse une condition de rectangle englobant de seuil.
     
    4. Procédé selon la revendication 3, comprenant en outre l'étape consistant à mettre en œuvre (306) une transformée de distance d'avant-plan sur l'image, et à supprimer (308) des contours dans l'image présentant une valeur de distance inférieure à un seuil de distance.
     
    5. Procédé selon la revendication 4, dans lequel un ou plusieurs éléments parmi le seuil de distance, la condition de rectangle englobant de seuil et la condition d'ajustement de distribution statistique de seuil, sont déterminés par un algorithme d'apprentissage.
     
    6. Procédé selon la revendication 1, dans lequel l'étape de test itératif de différentes combinaisons de sources de lumière (108) consiste à tester de manière itérative les différentes combinaisons de sources de lumière (108) selon un ordre qui est basé sur une occlusion prédite pour chacune d'une ou plusieurs combinaisons.
     
    7. Procédé selon la revendication 6, comprenant en outre l'étape consistant à détecter une pose de tête, et l'étape consistant à déterminer (208) une estimation de la combinaison de sources de lumière (108) qui produira la plus faible quantité d'occlusion sur la base de la pose de tête.
     
    8. Procédé selon la revendication 1 ou 2, comprenant en outre l'étape consistant à désigner un point saturé dans une image, en tant qu'un éblouissement, et l'étape consistant à mettre en correspondance l'éblouissement avec une source de lumière sélectionnée (108) dans une première combinaison de sources de lumière (108), sur la base d'une détermination de la présence ou non de l'éblouissement dans une image successive capturée pendant la projection d'une seconde combinaison distincte de sources de lumière (108) qui n'inclut pas la source de lumière sélectionnée (108).
     
    9. Procédé selon la revendication 1 ou 2, dans lequel une ou plusieurs sources de la pluralité de sources de lumière (108) sont orientées différemment les unes des autres, et dans lequel, lorsqu'elle dépend de la revendication 1, l'étape consistant à tester de manière itérative différentes combinaisons de sources de lumière (108) consiste à tester de manière itérative différentes combinaisons de sources de lumière (108) présentant différentes combinaisons d'orientations, ou dans lequel, lorsqu'elle dépend de la revendication 2, l'étape consistant à projeter de manière itérative de la lumière à partir de différentes combinaisons de sources de lumière (108) consiste à projeter de manière itérative de la lumière à partir de différentes combinaisons de sources de lumière (108) présentant différentes combinaisons d'orientations.
     
    10. Système de poursuite oculaire comprenant :

    une pluralité de sources de lumière (108) ;

    une caméra (119) configurée de manière à capturer une image d'une lumière provenant des sources de lumière, telle qu'elle est réfléchie par un œil d'un utilisateur ;

    un dispositif logique (702) ; et

    un dispositif de stockage (704) stockant des instructions exécutables par le dispositif logique (702) ;

    dans lequel le dispositif logique (702), lorsqu'il exécute les instructions du dispositif de stockage (704), est configuré de manière à mettre en œuvre un procédé selon l'une quelconque des revendications 1 à 9.


     




    Drawing



























    REFERENCES CITED IN THE DESCRIPTION



    This list of references cited by the applicant is for the reader's convenience only. It does not form part of the European patent document. Even though great care has been taken in compiling the references, errors or omissions cannot be excluded and the EPO disclaims all liability in this regard.

    Patent documents cited in the description