Gaze tracker and method for tracking a gaze direction

Abstract

Gaze tracker with: an imaging device (110) configured to obtain images of a user's eye using at least two cameras (111, 113) and at least two illuminations (115, 117) and a gaze direction calculator (120) configured to detect a reflection point of the illumination from each of the images of the eye and to extract a central point of a cornea, wherein the gaze direction calculator (120) is configured to detect a central point of the pupil from each of the images of the eye and to calculate a gaze direction vector using the central point of the cornea and the central point of the pupil, wherein, when at least two reflection points of the illumination from each of the images of the eye are detected, the gaze direction calculator (120) is configured to calculate vectors connecting each illumination and the corresponding reflection point of the illumination and to detect an intersection point between the calculated vectors as the central point of the cornea, and wherein, when only one reflection point of the illumination from one or more images of the eye among the images of the eye is detected, the gaze direction calculator (120) is configured to extract the central point of the cornea using data regarding a distance between a previously recorded reflection point of the illumination and the central point of the cornea and the one reflection point of the illumination, characterized in that the gaze direction calculator (120) is configured to calculate distances from the central point of the cornea to the at least two reflection points of the illumination and to accumulate and record the calculated data regarding the distance.

Description

Technical area

[0001] The present disclosure relates to a gaze tracking device and a gaze tracking method for tracking the gaze direction of a user. TECHNICAL BACKGROUND

[0002] The statements in this section merely constitute background information related to the present disclosure and do not necessarily constitute prior art.

[0003] Gaze tracking is used in various fields, such as monitoring a driver's state, detecting intent, and controlling a vehicle device that uses gaze tracking. Recently, gaze tracking has been applied to monitor a driver's lapse of attention.

[0004] Such a gaze tracker captures an image of an eye using a stereo camera and detects two illumination reflection points from the captured image of the eye. Additionally, the gaze tracker calculates a corneal central point using the two detected illumination reflection points and calculates a user's gaze direction using the calculated corneal central point and a pupil central point.

[0005] However, we discovered that an ordinary gaze tracker is able to track a gaze direction only when two reflection points of illumination are generated in an iris, and therefore, when only one reflection point of illumination is generated in an iris due to face panning, it is not possible to track a gaze direction, which narrowly and excessively restricts a gaze tracking range.

[0006] US 2014 / 0 313 308 A1 is prior art which discloses a device for tracking a line of sight and forms the preamble of the independent claims. Description of the invention

[0007] The present disclosure provides a gaze tracker and a gaze tracking method for tracking a user's gaze even when only one reflection point is formed at a cornea during gaze tracking using a stereo camera having two cameras and two illuminations.

[0008] The invention is defined by the independent claims.

[0009] A gaze tracker comprises an imaging device configured to obtain images of a user's eye using two or more cameras and two or more illuminations, and a gaze calculator configured to detect a reflection point of the illumination from each image of an eye, to extract a central point of a cornea, to detect a central point of a pupil from each image of an eye, and to calculate a gaze direction vector using the central point of the cornea and the central point of the pupil, wherein, when two or more reflection points of the illumination are detected from each of the images of the eye, the gaze calculator calculates vectors connecting each illumination and a corresponding reflection point of the illumination and detects an intersection point between the calculated vectors as the central point of the cornea, and,if only one reflection point of the illumination is detected from one or more images of the eye among the images of the eye, the gaze direction calculator extracts the central point of the cornea using distance data between a previously stored reflection point of the illumination and the central point of the cornea and the one reflection point of the illumination.

[0010] The gaze direction calculator calculates distances from the central point of the cornea to the two or more reflection points of the illumination and collects and records the calculated distance data.

[0011] The gaze direction calculator can convert two-dimensional coordinates of the reflection point of the illumination and the central point of the pupil into three-dimensional coordinates.

[0012] The gaze direction calculator may determine an illumination that forms the one reflection point of the illumination on a cornea among the two or more illuminations using a face direction vector.

[0013] The gaze direction calculator can calculate a similarity between reflection points of illumination in an image of an eye from which two or more reflection points of illumination are detected and an image of an eye from which one reflection point of illumination is detected, among the images of the eye, and determine an illumination which forms the one reflection point of illumination on a cornea.

[0014] The gaze direction calculator can calculate a vector connecting between the one reflection point of the illumination and the specific illumination and extract the central point of the cornea using the calculated vector and the distance data.

[0015] The distance data may be obtained by calculating an average distance from distance data for a predetermined recent time period among data points of distance data from a previously recorded central point of the cornea to a reflection point of the illumination.

[0016] A method for tracking a gaze direction of a gaze tracker having two or more cameras and two or more illuminations comprises obtaining images of a user's eye using the two or more cameras, detecting a reflection point of illumination from each of the images of the eye, after detecting two or more reflection points of illumination from each of the images of the eye, calculating vectors connecting each illumination to a reflection point of illumination corresponding thereto, and detecting an intersection point between the calculated vectors as a central point of a cornea, wherein, if there is only one reflection point of illumination from one or more images of an eye among the images of the eye,the central point of the cornea is detected using distance data between a previously recorded reflection point of the illumination and the central point of the cornea and the one reflection point of the illumination, after detecting the central point of the cornea, extracting the central point of a pupil from the image of the eye and calculating a gaze direction vector using the central point of the cornea and the central point of the pupil.

[0017] Detecting the reflection point of the illumination may include converting two-dimensional coordinates of the reflection point of the illumination into three-dimensional coordinates.

[0018] Detecting the intersection point of the vectors as the central point of the cornea comprises calculating distances from the central point of the cornea to the two or more reflection points of the illumination and accumulating and recording the calculated distance data.

[0019] Detecting the central point of the cornea using the one reflection point of the illumination may include determining an illumination corresponding to the one reflection point of the illumination, calculating a first vector for connecting the one reflection point of the illumination and the determined illumination, and extracting the central point of the cornea using the first vector and the distance data.

[0020] Determining the illumination corresponding to the one reflection point of the illumination may include determining an illumination constituting the one reflection point of the illumination on a cornea among the two or more illuminations using a vector of a face direction.

[0021] Determining the illumination corresponding to the one reflection point of the illumination may include calculating a similarity between reflection points of the illumination in an image of an eye from which two or more reflection points of the illumination are detected and an image of an eye from which one reflection point of the illumination is detected, and determining an illumination constituting the one reflection point of the illumination on a cornea.

[0022] The distance data can be obtained by calculating an average distance from distance data for a predetermined recent time period among the entries of distance data from a previously recorded central point of a cornea to a reflection point of the illumination.

[0023] Further areas of applicability will become apparent from the description provided herein. It should be understood that the description and specific examples are for purposes of illustration only and are not intended to limit the scope of the present disclosure. Short description of the drawings

[0024] In order that the content of the disclosure may be well understood, various forms of it will now be described, which are intended only as examples, with reference to the accompanying drawings in which Fig.1 is a block diagram of a gaze tracker, Fig. 2 is a diagram for explaining the extraction of a central point of a cornea and Fig. 3A and Fig. 3B are flowcharts illustrating a method for gaze tracking.

[0025] The drawings described herein are for illustration purposes only and are not intended to limit the scope of the present disclosure in any way. Detailed description

[0026] The following description is provided by way of example only and is not intended to limit the present disclosure, application, or uses. It should be noted that throughout the drawings, corresponding reference characters refer to like or corresponding parts and features.

[0027] It should also be noted that the terms “comprises” and / or “comprising”, when used in this specification, indicate the presence of the designated components but do not exclude the presence or addition of other components.

[0028] Also, terms such as "unit" or "module," etc., should be understood as a unit that performs at least one function or activity, and which may be implemented as hardware, software, or a combination of hardware and software. In this context, the singular forms "a," "an," and "the" are intended to include the plural unless the context clearly indicates otherwise.

[0029] A gaze tracker according to the present disclosure has two or more cameras and two or more illuminators, but an example in which the gaze tracker has two cameras and two illuminators will be described for understanding the present disclosure.

[0030] Fig. 1 is a block diagram of a gaze tracker.

[0031] As in Fig. 1, the gaze tracker may include an imaging device 110 and a gaze calculator 120.

[0032] The imaging device 110 may be a stereo camera and may capture images of a user's left and right eyes. The imaging device 110 may include a first camera 111, a second camera 113, a first illuminator 115, and a second illuminator 117.

[0033] The first camera 111 and the second camera 113 can simultaneously photograph the same object (a user's eye) from different positions. In other words, the imaging device 110 can capture two images of an eye, respectively, with respect to the user's left and right eyes. When the images are generated by the first camera 111 and the second camera 113, the first illuminator 115 and the second illuminator 117 can emit infrared rays to the user's eyes.

[0034] The first camera 111 and the second camera 113 may be implemented as an infrared camera with an infrared filter. Additionally, the first illumination 115 and the second illumination 117 may be implemented as a light-emitting diode (LED), a lamp, or the like for emitting infrared rays.

[0035] Although the imaging device 110 photographs an eye of a user, the present disclosure is not limited thereto, and the imaging device 110 may be configured to photograph a face of a user and extract a region of an eye from the image of the face taken by an image processing device.

[0036] The gaze direction calculator 120 may calculate the gaze direction of a user from images of the eye acquired by the imaging device 110. The gaze direction calculator 120 may include an illumination reflection point detection module 121, a cornea central point extraction module 123, a pupil central point extraction module 125, and a gaze direction extraction module 127.

[0037] The illumination reflection point detection module 121 can detect an illumination reflection point generated by the first illumination 115 and / or the second illumination 117 from each image of an eye taken by the first camera 111 and the second camera 113. In other words, the illumination reflection point detection module 121 can detect the illumination reflection point generated by the first illumination 115 and / or the second illumination 117 from the image of the eye taken by the first camera 111. In addition, the illumination reflection point detection module 121 can detect the illumination reflection point generated by the first illumination 115 and / or the second illumination 117 from the image of the eye taken by the second camera 113.

[0038] The illumination reflection point detection module 121 may convert the two-dimensional coordinates of the detected illumination reflection points into three-dimensional coordinates using 3D reconstruction techniques such as triangulation. For example, coordinates of a first illumination reflection point generated by the first illumination 115 may be extracted from a first image of an eye and a second image of an eye captured by the first camera 111 and the second camera 113, and three-dimensional coordinates of the first illumination reflection point may be calculated using the coordinates of the first illumination reflection point extracted from the first image of the eye and the coordinates of the first illumination reflection point extracted from the second image of the eye.

[0039] In addition, the illumination reflection point detection module 121 may detect an illumination reflection point using a high-pass filter.

[0040] With reference to Fig. 2, when the current position coordinates of the first illumination 115 and the second illumination 117 are L1 and L2, respectively, the illumination reflection point detection module 121 can calculate the three-dimensional coordinates L1' and L2' of the first illumination reflection point and the second illumination reflection point formed on the cornea by the first illumination 115 and the second illumination 117.

[0041] The corneal central point extraction module 123 may calculate a corneal central point using the detected illumination reflection point (the first illumination reflection point and / or the second illumination reflection point) and position coordinates of the illuminators 115 and / or 117 corresponding to the detected illumination reflection point. In other words, the corneal central point extraction module 123 may calculate a first vector (straight line D3) connecting a three-dimensional coordinate L1' of the first illumination reflection point and a current position coordinate L1 of the first illumination 115.In addition, the corneal central point extraction module 123 may calculate a second vector (straight line D4) connecting a three-dimensional coordinate L2' of the second illumination reflection point and a current position coordinate L2 of the second illumination 117. The corneal central point extraction module 123 may extract an intersection point between the first vector and the second vector as a corneal central point O.

[0042] The corneal central point extraction module 123 may extract the corneal central point O and then calculate distances D1 and D2 to the first illumination reflection point L1' and to the second illumination reflection point L2' from the corneal central point O. In addition, the corneal central point extraction module 123 may accumulate and record the calculated distance data in a recording means (not shown). Here, the distance data may be recorded in the form of Table 1 shown below. (Table 1) Frame # Image of the right eye Image of the left eye D1 D2 D1 D2 Frame 1 22mm 21mm 20mm 17mm Frame 2 25mm 25mm 22mm 24mm • • • • • • • • • •

[0043] After detecting an illumination reflection point, the corneal central point extraction module 123 may calculate an average distance from distance data for a predetermined recent period among the distance data between the corneal central point and the first illumination reflection point detected from previous images of the eye previously recorded in a recording means (not shown). For example, the corneal central point extraction module 123 may calculate an average of the distances between the corneal central point and the first illumination reflection point from five recent frames.

[0044] In addition, the corneal central point extraction module 123 may calculate an average distance from distance data for a predetermined recent period from distance data between the corneal central point and the second reflection point of illumination obtained from previous images of the eye previously recorded in a recording means (not shown).

[0045] Although it has been described that an average distance is calculated and used from distance data for a predetermined recent period of time from previously recorded distance data, the present disclosure is not limited thereto, and the most recent distance data entry of the previously recorded distance data may be used.

[0046] When only one illumination reflection point is detected by the illumination reflection point detection module 121, the cornea central point extraction module 123 may check an illumination that constitutes the detected illumination reflection point on the cornea among the first illumination 115 and the second illumination 117.

[0047] In this case, the corneal central point extraction module 123 may determine whether the detected illumination reflection point is the first illumination reflection point formed on the cornea by the first illumination 115 or the second illumination reflection point formed on the cornea by the second illumination 117 using a face direction vector. The face direction vector may be calculated from an image of a user's face obtained by the imaging device 110.

[0048] The corneal central point extraction module 123 may calculate a similarity between illumination reflection points in an image of an eye from which two or more illumination reflection points are detected and an image of an eye from which one illumination reflection point is detected, and determine an illumination constituting the detected one illumination reflection point on the cornea under the first illumination 115 and the second illumination 117.

[0049] For example, when the first reflection point of illumination is positioned away from a central point of the pupil to the left by 3 pixels and the second reflection point of illumination is positioned away from the central point of the pupil to the right by 3 pixels in an image of an eye from which the two reflection points of illumination are detected, a distance between the one detected reflection point of illumination and the central point of the pupil is calculated, and whether the one detected reflection point of illumination is positioned to the left or to the right of the central point of the pupil can be determined based on the calculated distance.

[0050] The corneal central point extraction module 123 may calculate a third vector connecting the first illumination reflection point to the first illumination 115 if the detected illumination reflection point is the first illumination reflection point. Additionally, the corneal central point extraction module 123 may calculate a corneal central point using the third vector and an average distance between the first illumination reflection point and the corneal central point.

[0051] If the detected illumination reflection point is the second illumination reflection point, the corneal central point extraction module 123 may calculate the third vector connecting the second illumination reflection point and the second illumination 117. Additionally, the corneal central point extraction module 123 may calculate the corneal central point using the third vector and an average distance between the second illumination reflection point and the corneal central point.

[0052] The pupil central point extraction module 125 may perform image processing on an image of an eye to determine the pupil central point. In this case, the pupil central point extraction module 125 may capture a region of an eye from the image of an eye acquired by the imaging device 110 and capture the pupil central point from the captured region of the eye. The pupil central point extraction module 125 may reconstruct the two-dimensional pupil central point captured from the eye image into a three-dimensional pupil central point.

[0053] The gaze direction calculation module 127 may calculate a gaze direction vector along which a user gazes using the central point of the pupil and the central point of the cornea. That is, the gaze direction calculation module 127 may calculate a straight line between the central point of the pupil and the central point of the cornea as the gaze direction vector.

[0054] Fig. 3A and Fig. 3B are flowcharts illustrating a method for tracking a gaze direction.

[0055] As in Fig.As shown in Figure 3A, the gaze calculator 120 of the gaze tracker may obtain an image of a user's eye through the imaging device 110 (S101). In this case, the gaze calculator 120 may capture the image of the user's eye through the first camera 111 and the second camera 113. In other words, the gaze tracker may obtain two images, respectively, relating to the user's right and left eyes.

[0056] The gaze tracker's illumination reflection point detection module 121 may detect one or more illumination reflection points from the eye images output by the first camera 111 and the second camera 113 (S103). The illumination reflection point detection module 121 may convert the two-dimensional coordinates of the detected illumination reflection points into three-dimensional coordinates using 3D reconstruction techniques such as triangulation.

[0057] The corneal central point extraction module 123 of the gaze tracker may check whether the two illumination reflection points are detected by the illumination reflection point detection module 121 (S105). In other words, the corneal central point extraction module 123 may check whether the first illumination reflection point and the second illumination reflection point formed on the cornea by the first illumination 115 and the second illumination 117 are detected.

[0058] When the two illumination reflection points are detected from the respective images of the eye, the corneal central point extraction module 123 may calculate vectors connecting the respective detected illumination reflection points to the corresponding illuminations 115 and 117 (S107). In other words, the corneal central point extraction module 123 may calculate a first vector connecting a three-dimensional coordinate of the first illumination reflection point and an actual position coordinate of the first illumination 115, and a second vector connecting a three-dimensional coordinate of the second illumination reflection point and an actual position coordinate of the second illumination 117.

[0059] The corneal central point extraction module 123 may extract a point where the calculated vectors intersect with each other as the corneal central point (S109). For example, the corneal central point extraction module 123 may extract the intersection point between the previously calculated first vector and second vector as the corneal central point.

[0060] The corneal central point extraction module 123 may calculate and record distances from the corneal central point to the first illumination reflection point and the second illumination reflection point (S110).

[0061] When extracting the corneal central point, the gaze tracker may extract the pupil central point from the eye image through image processing (S111). In this case, the pupil central point extraction module 125 may acquire a region of an eye from the image obtained by the imaging device 110 and acquire the pupil central point from the eye region. In this case, the pupil central point extraction module 125 may reconstruct a two-dimensional pupil central point into a three-dimensional pupil central point. Although the pupil central point is extracted after the corneal central point is extracted, the present disclosure is not limited to this, and the pupil central point and the corneal central point may also be acquired simultaneously.

[0062] The gaze direction calculation module 127 of the gaze tracker may calculate a straight line connecting the central point of the cornea and the central point of the pupil as a gaze direction vector (S113).

[0063] If two illumination reflection points are not detected at step S105, the gaze tracker may check whether the number of detected illumination reflection points is one (S121). The gaze tracker may check whether only one illumination reflection point is detected from one or more images of the eye among the images of the eye captured by the first camera 111 and the second camera 113.

[0064] If only one illumination reflection point is detected, the gaze tracker may calculate an average distance between a previously recorded illumination reflection point and the corneal central point (S123). The corneal central point extraction module 123 may calculate an average distance between the first illumination reflection point and the corneal central point and an average distance between the second illumination reflection point and the corneal central point for a predetermined recent period among entries regarding the distance data to the first illumination reflection point and the second illumination reflection point from a previously recorded corneal central point.

[0065] The gaze tracker may determine an illumination corresponding to the detected illumination reflection point (S125). In this case, the gaze tracker may check an illumination that forms the detected illumination reflection point among the first illumination 115 and the second illumination 117 using a face direction vector.

[0066] The gaze tracker may calculate a third vector connecting the detected illumination reflection point and a corresponding illumination (S127). In other words, if the detected illumination reflection point is the first illumination reflection point, the corneal central point extraction module 123 may calculate a vector connecting the first illumination reflection point and the first illumination 115. If the detected illumination reflection point is the second illumination reflection point, the corneal central point extraction module 123 may calculate a vector connecting the second illumination 117 and the second illumination reflection point.

[0067] The gaze tracker can calculate the central point of the cornea using the third vector and the average distance (S129).

[0068] Then, the gaze tracker can extract the central point of the pupil (S111) and can calculate a gaze direction vector using the central point of the cornea and the central point of the pupil (S113).

[0069] Although the case described so far is that S125 was performed after S123 was performed, S123 can also be performed after S125 was performed.

[0070] For example, if the one detected reflection point of illumination is the first reflection point of illumination, the corneal central point extraction module 123 may calculate an average distance from the distance data for a predetermined recent period among entries regarding the distance data between the corneal central point and the first reflection point of illumination acquired from previous images of eyes previously stored in a recording means (not shown).

[0071] When the one detected illumination reflection point is the second illumination reflection point, the corneal central point extraction module 123 may calculate an average distance from distance data for a predetermined recent period among entries regarding the data on distances between the corneal central point and the second illumination reflection point acquired from previous images of eyes previously stored in a recording device (not shown).

[0072] When only one reflection point of illumination is detected from an image of an eye, an illumination that generates the detected reflection point of illumination may be determined using a vector of a facial direction, but the present disclosure is not limited thereto.

[0073] For example, when one reflection point of illumination is detected from one image of one eye and two reflection points of illumination are detected from the other image of one eye among two images of one eye obtained by two cameras, the similarity between the reflection points of illumination in the two images of one eye can be calculated and an illumination that produces the one detected reflection point of illumination can be determined.

[0074] According to the present disclosure, when a user's gaze direction is tracked using a stereo camera with two cameras and two illuminations, even if only one illumination reflection point is formed on the cornea, a user's gaze direction can be tracked using only one illumination reflection point. Accordingly, a gaze tracking range of the gaze tracker can be enlarged. Symbol of each of the elements in the figures 110 Imaging device 111 first camera 113 second camera 115 first lighting 117 second lighting 120 viewing direction calculators 121 Detection module for the reflection point of the lighting 123 Extraction module for the central point of the cornea 125 Extraction module for the central point of the pupil 127 Calculation module for the viewing direction S101 Obtaining the image of a user's eye by two or more cameras S103 Detecting the reflection point of the illumination of each image of an eye S105 Are two reflection points of the lighting detected? S107 Calculating a first vector connecting a first reflection point of the illumination and the first illumination and a second vector connecting the second reflection point of the illumination and the second illumination S109 Extracting an intersection point between the first vector and the second vector as the central point of the cornea S110 Calculating and recording distances from the central point of the cornea to the first reflection point of the illumination and the second reflection point of the illumination S111 Extracting the central point of the pupil S113 Calculate the gaze direction vector using the central point of the cornea and the central point of the pupil S121 Is a reflection point of the lighting detected? S123 Calculate the average distance between the previously recorded reflection point of the illumination and the central point of the cornea S125 Determine the illumination according to the detected reflection point of the illumination S127 Calculating the third vector connecting the detected reflection point of the illumination and the corresponding illumination S129 Detecting the central point of the cornea using the third vector and the average distance

Claims

[1] Gaze tracker with: an imaging device (110) configured to obtain images of a user's eye using at least two cameras (111, 113) and at least two illuminations (115, 117) and a gaze direction calculator (120) configured to detect a reflection point of the illumination from each of the images of the eye and to extract a central point of a cornea, wherein the gaze direction calculator (120) is configured to detect a central point of the pupil from each of the images of the eye and to calculate a gaze direction vector using the central point of the cornea and the central point of the pupil, wherein, when at least two reflection points of the illumination from each of the images of the eye are detected, the gaze direction calculator (120) is configured to calculate vectors connecting each illumination and the corresponding reflection point of the illumination and to detect an intersection point between the calculated vectors as the central point of the cornea, and wherein, when only one reflection point of the illumination from one or more images of the eye among the images of the eye is detected, the gaze direction calculator (120) is configured to extract the central point of the cornea using data regarding a distance between a previously recorded reflection point of the illumination and the central point of the cornea and the one reflection point of the illumination, characterized by , that the gaze direction calculator (120) is configured to calculate distances from the central point of the cornea to the at least two reflection points of the illumination and to accumulate and record the calculated data regarding the distance. [2] The gaze tracker according to claim 1, wherein the gaze calculator (120) is configured to convert two-dimensional coordinates of the reflection point of the illumination and the central point of the pupil into three-dimensional coordinates. [3] A gaze tracker according to any one of the preceding claims, wherein the gaze direction calculator (120) is configured to determine an illumination forming the one reflection point of the illumination on a cornea among the at least two illuminations (115, 117) using a vector of a facial direction. [4] A gaze tracker according to any one of the preceding claims, wherein the gaze calculator (120) is configured to calculate a similarity between reflection points of illumination in the image of the eye from which the at least two reflection points of illumination are detected and the image of the eye from which one reflection point of illumination is detected, among the images of the eye, and to determine an illumination that forms the one reflection point of illumination on the cornea. [5] The gaze tracker of claim 3, wherein the gaze calculator (120) is configured to calculate a vector connecting the one reflection point of the illumination and the determined illumination and to extract the central point of the cornea using the calculated vector and distance data. [6] A gaze tracker according to any one of the preceding claims, wherein the gaze calculator (120) is configured to obtain the distance data by calculating an average distance from distance data for a predetermined recent period among entries relating to the data on distances from a previously recorded central point of the cornea to a reflection point of the illumination. [7] A method for tracking a line of sight from a line of sight tracker having at least two cameras (111, 113) and at least two illuminations (115, 117), the method comprising: capturing images of a user's eye using the at least two cameras (111, 113), capturing a reflection point of the illumination from each of the images of the eyes, upon detecting at least two reflection points of the illumination from each of the images of the eyes, calculating vectors connecting each illumination to a corresponding reflection point of the illumination and detecting an intersection point between the calculated vectors as a central point of the cornea, and, if only one reflection point of the illumination is detected from one or more images of the eye among the images of the eyes, detecting the central point of the cornea using distance data between a previously recorded reflection point of the illumination and the central point of the cornea and the one reflection point of the illumination, when the central point of the cornea has been detected, extracting a central point of the pupil from the image of the eye and calculating a gaze direction vector using the central point of the cornea and the central point of the pupil, characterized by that detecting the intersection point of the vectors as the central point of the cornea comprises calculating distances from the central point of the cornea to the at least two reflection points of the illumination and accumulating and recording calculated distance data. [8] The method of claim 7, wherein detecting the reflection point of the illumination comprises converting two-dimensional coordinates of the reflection point of the illumination into three-dimensional coordinates. [9] The method of claim 7 or 8, wherein detecting the central point of the cornea using the one reflection point of the illumination comprises: Determining an illumination corresponding to the one reflection point of the illumination, Calculating a first vector connecting the one reflection point of the illumination and the determined illumination and Extract the central point of the cornea using the first vector and the distance data. [10] The method of claim 9, wherein determining the illumination corresponding to the one reflection point of the illumination comprises determining an illumination forming the one reflection point of the illumination at the cornea among the at least two illuminations (115, 117) using a vector of a facial direction. [11] The method of claim 9 or 10, wherein determining the illumination corresponding to the one reflection point of the illumination comprises calculating the similarity between reflection points of the illuminations in an image of an eye from which two or more reflection points of the illumination are detected and an image of an eye from which one illumination is detected, and determining an illumination which forms the one reflection point of the illumination on the cornea. [12] A method according to any one of claims 7 to 11, wherein the data on distance is obtained by calculating an average distance from data on distance for a predetermined recent period among entries on the data on distance from a previously recorded central point of the cornea to the reflection point of the illumination.

Images