JPS62176427A - Apparatus for measuring position of notice point - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a method and apparatus for correcting distortion of an eye mark signal obtained from an eye camera.

(Prior technology) An eye camera is used to analyze the distribution of gaze points on objects in the surrounding space. It irradiates the protrusion of the eyeball with near-infrared light and uses an imaging device to determine the position of the reflected light in the X and Y directions. By detecting it, the position coordinates of the gaze point are obtained as an eye mark signal, and the eye mark signal relative to the direction of the head is output as a digital value, and the field of view taken and displayed by a camera fixed to the head is output. This is a device that displays a light spot indicating an eye mark signal within an image.

However, the eye mark signal contains non-negligible distortions due to differences in the size and shape of each individual's eyeball, the radius of curvature of the protruding part of the eyeball, and errors in the installation of the eye camera. Specifically, this phenomenon occurs in which the position of the image of the object of gaze and the position of the light spot indicating the eye mark signal are shifted in the visual field image. Therefore, in order to correct this distortion, conventionally,
A method is adopted in which the electrical magnification of the eye mark signal in each of the X and Y directions is adjusted.

(Problem to be Solved by the Invention) However, the above method can only correct the linear component of the distortion of the eye mark signal, and is difficult to correct in practice, especially in the periphery of the visual field image where the distortion increases non-linearly. Not enough. The present invention measures the distortion of the eye mark signal in advance, records it as a correction table, and refers to this at the time of measurement to correct the linear and nonlinear components of the distortion, thereby obtaining a highly accurate gaze point position measuring device. It is an object.

(Means for Solving the Problems) According to the present invention, an eye camera, an image input means for digitizing a video signal output from the eye camera, and a target group that is a feature point in a visual field image are provided. A feature point setting means installed in the surrounding space, a coordinate detection means for detecting the coordinates of the feature point from the digitized visual field image, and a coordinate detection means indicating the coordinates of the gaze point in the visual field image when the target group is sequentially gazed. table storage means for storing in advance the eye mark signal and the coordinates as a feature point of the gazed target detected by the coordinate detection means in association with each other as an eye mark table and a feature point table;
By having a distortion correction means that corrects the distortion of the eye mark signal obtained during measurement using the value of the feature point table corresponding to the value of one or more eye mark tables near the coordinate value of the eye mark signal, It is possible to reduce the distortion of the eye mark signal and obtain a highly accurate gaze point position measuring device.

(Function) The present invention solves the conventional problems by adopting the configuration described in detail. The operating principle of the present invention will be briefly explained below.

Prior to measurement, multiple targets are prepared in the surrounding space that will serve as feature points in the visual field image. The person to be measured sets the eye camera on his head in the state at the time of measurement and gazes at these targets in sequence. The coordinates of the gazed target as a feature point in the visual field image and the coordinates of the eye mark signal at the time of gaze are associated with each other and sequentially stored in the table storage means (hereinafter, each stored coordinate will be referred to as a feature point table value). , called the eyemark table value). During measurement, the stored correction table is used to compare the eye mark signal sent from the eye camera with the correction table, search for one or more eye mark table values that are close to the eye mark signal, and then A correction value is calculated using the feature point table value stored corresponding to the selected eye mark table value.

FIG. 2 is a schematic diagram illustrating the operating principle of the present invention when targets are arranged, for example, in a grid. In the figure, X,
The Y plane is the field of view image plane of the camera, and each lattice point of the lattice on the plane indicated by O indicates an eye mark signal (eye mark table value) when gazing at a target. The Z axis indicates the component in the X direction of the coordinates (feature point table values, indicated by .) in the visual field image of the target corresponding to each eye mark signal.

Suppose now that an eye mark signal is input at a position shown by P in FIG. 2 at a certain point during measurement. For example, El, E2. If the three eye mark table values of E3 are selected, the table is further searched for the X coordinates Cxl, Cx2°Cx3 of the feature point table values corresponding to each point. The correction value Px of P in the X direction is Cx
l, Cx2. Calculated using Cx3. The same holds true for the Y direction.

(Example) FIG. 1 is a block diagram showing one embodiment of the present invention.

The feature point setting means 10 sets a target object that will become a feature point in the visual field image in the surrounding space when creating the distortion correction table. The eye camera 11 sends a visual field image to an image input means 12 via a signal line 100, a display device 16 such as a CRT display via a signal line 109, and an eye mark signal to a control device 17 via a signal line 101. send to The image input means 12 converts the sent visual field image into a digital image and sends it to the coordinate detection means 13 via the signal line 102. The coordinate detection means 13 detects the feature points of the sent digitized visual field image and sends the coordinate values to the table storage means 14 via the signal line 104. The table storage means 14 sequentially stores the coordinate values of the sent feature points and the coordinate values of the eye mark signal sent from the control device 17 via the signal line 103 in association with each other. During measurement, the distortion correction means 15 receives and receives an eye mark signal from the control device 17 via the signal line 103, and reads the eye mark table value in the vicinity of the eye mark signal from the signal line 105 from the correction table stored in the table storage means 14. 106, and the feature point table value corresponding to the eyemark table value is obtained via the signal line 106. Further, the distortion correction means 15 uses the obtained feature point table values to calculate a correction value, for example, by a method described later, and sends it to the control device 17 via the signal line 107. The control device 17 sends the correction value to the display device 16 via a signal line 108. The display device 16 increases the display brightness of the position on the screen corresponding to the sent coordinate value and displays it as a light spot. Furthermore, the control device 17 via a signal line 110,
The above operations are controlled throughout the device.

Taking as an example a case where a CRT display and light spots arranged in a grid pattern on the screen are used as the feature point setting means, the operation of this embodiment will be explained separately during correction table creation and gaze point measurement.

When creating the correction table, the control device 17
On the RT display, light points that are feature points with particularly high brightness in the visual field image are displayed one by one at positions corresponding to each grid point of the grid on the screen of the CRT display. The person to be measured sets the eye camera 11 in the state at the time of measurement, fixes his head as much as possible, and gazes at the displayed light spot. The eye camera 11 sends a visual field image to the image input means 12 and an eye mark signal to the control device 17. The visual field image is transmitted to the image input means 12.
The data is AD converted and sent to the coordinate detection means 13. The coordinate detection means 13 binarizes the sent visual field image using a luminance level lower than the image of the feature point and higher than other image areas as a threshold value. Furthermore, the center of gravity of the feature points extracted by binarization is calculated and sent to the table storage means 14. Also, at this time, the control device 17 sends an eye mark signal to the table storage means 14 when the user gazes at a light spot on the screen of the CRT display. The table storage means 14 stores the coordinates of the center of gravity of the sent feature points as a feature point table value and the eye mark signal as an eye mark table value in association with each other.

The above operations are performed for all grid points to create a correction table.

When measuring the gaze point, the eye camera 11 sends an eye mark signal to the control device 17 and a field of view image to the display device 16, which is constituted by, for example, a monitor. The control device 17 sends the sent eye mark signal to the distortion correction means 15. The distortion correction means 15 compares the coordinates of the sent eyemark signal with the eyemark table values stored in the table storage means 14, and searches for, for example, three closest eyemark table values. In other words, assuming that the eye mark signal is (EX, EY) and the eye mark table value is (CXi, CYi), the calculation expressed by equation (1) is executed for all eye mark table values, and the eye mark table values are Select 3 mark table values.

L=(EX-CXi)2+(EY-CYi)2(1) Furthermore, the correction means 15 stores three feature point table values corresponding to the selected three eyemark table values from the table storage means 14. A correction value is calculated using the coordinate values of these three eye mark table values and three feature point table values. FIG. 3 is a schematic diagram showing a calculation method for the correction value in the X direction. (The same applies to the Y direction.) In the figure, El, E2. E3 has the eye mark table values of the three selected points in the X and Y coordinates, and the points where the two coordinates are 0, C1, C2, and C3 are each E1.
~E3 have the same X and Y coordinates, and El, E2. E
The X coordinate values tl, t2 . of the feature point table values corresponding to 3.
A point having t3 as two coordinates, P indicates a point having eye-marker signals on the X and Y coordinates, and having two coordinates as 0. At this time, the correction value Px of P is that the perpendicular lines drawn from P are C1, C2, and C3.
It is expressed by the two coordinates of the point that intersects the plane passing through. That is, (
2) Just solve the equation expressed by the formula for 2. However, in equation 2), ITI indicates the determinant of T.

The control device 17 receives the correction value of the eye mark signal obtained by the above method from the distortion correction means 15, and displays its coordinate position on the display device 16, for example, as a light spot on a visual field image.

Furthermore, by creating a correction table using the method described below, it is possible to simplify the method of searching for the eye mark table value and the method of calculating the correction value. When creating the correction table, the control device 17 presets a group of coordinate values arranged in a grid in the visual field image. CRT used as the feature point setting means 10 as a gaze target
On the display, a single light spot that is sufficiently high in the heel skin is displayed. The control device 17 moves the light spot on the screen at a speed that the subject can follow with his or her line of sight. The control device 17 monitors the eye mark signal at this time and determines whether it is equal to one point out of a preset coordinate value group.
Alternatively, move the light spot until the values are close enough. The visual field images at the time when they are equal or sufficiently close are captured by the image input means 12, and the center of gravity of the feature point is calculated by the coordinate detection means 13, and the center of gravity of the feature point is calculated by the coordinate detection means 13.
Sent to 4. The above operations are performed for all coordinate value groups set in advance. The feature point table values of the correction table obtained through the above operations are stored in such an order that the setting values of the respective corresponding eye mark signals are arranged in the order shown in FIG. 4, for example. In FIG. 4, numeral 41 schematically represents an eye mark signal set in advance in a grid pattern, and numeral 42 represents an arrow indicating the recording order of the corresponding feature point table values in the correction table. Since the eye mark table values are known in advance, there is no need to record them in the correction table.

For example, as shown in FIG. 5, the eye mark signal is X.

A method of searching the feature point table values will be described below when the lattice of MXN is set in 10 increments in both the Y direction. The eye mark signal at a certain point is P (EX, EY)
TE, EX, and EY are represented by a, b, c, and d, respectively, so that they remain in equation (3). However, d is a one-digit integer.

EX=10Xa+b EY=10Xc+d At this time, the feature point table value corresponding to the point indicated by a in the figure is stored in the aXn+bth position in the correction table. When b + d is less than 10, it means that P enters the triangular area like O in the figure, and the three points to be searched for are aXn+b indicated by A in Figure 5, and aXn+b indicated by Eng.
+1, which is the aXn+2Xb-th point indicated by A. b ten d
When is 10 or more, it means that P has entered the region like the force in the figure, so instead of the aXn+b-th point above, the three points to be searched are aXn+2Xb+1 shown in the figure.
The second point and the other two points.

FIG. 6 schematically represents the method described above (the method of calculating the correction value in the X direction using the three retrieved points).

The calculation method will be explained below, taking as an example the case where the eye mark signal P at a certain point of time falls into a triangular region like O in FIG. In FIG. 6, C1, C2, and C3 represent the X coordinate values of the feature point table values corresponding to the points A, B, and G in FIG.
It shows the point that has as. Q is X.

The Y coordinate in equation (3) indicates a certain point. At this time, the correction value Px of P in the It is indicated by the two coordinate values of the intersection point (Qx).In other words, the correction value can be obtained by the following calculation.If P is in the area of the fifth force, t2 is the force in Figure 5, and t3 is the force in Figure 5. In addition, the X coordinate of the feature point table value corresponding to each point A in the figure is 10-
Instead of b and d, 10-d may be substituted into the fourth equation.

(Effects of the Invention) As is clear from the above explanation, according to the gaze point position measuring device of the present invention, linear components and nonlinear components of distortion in gaze point coordinate output can be corrected by referring to the table. , it is possible to measure the position of the gaze point with high accuracy.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of the gaze point position measuring method of the present invention, FIG. 2 is a schematic diagram showing the operating principle of the present invention,
FIG. 3 is a schematic diagram showing an example of the correction value calculation method, FIG. 4 is a schematic diagram showing the storage order of data in the correction table,
FIG. 5 is a schematic diagram showing a feature point table value search method, and FIG. 6 is a schematic diagram showing another embodiment of a correction value calculation method. In the figure, 10 is a feature point setting means, 11 is an eye camera, 12 is an image inputting means, 13 is a coordinate detection means, 14 is a table storage means, 15 is a distortion correction means, 16 is a display device, and 17 is a Hi-1
Bi Hi 2 Hi 3

Claims (1)

[Claims] an eye camera, an image input means for digitizing a video signal output from the eye camera, a feature point setting means for setting a group of targets to be feature points in a visual field image in surrounding space; a coordinate detection means for detecting the coordinates of a feature point from a visual field image; and an eye mark signal indicating the coordinates of a gaze point in the visual field image when the target group is sequentially gazed; A table storage means for storing in advance the coordinates of a target feature point in association with an eye mark table and a feature point table, respectively; A gaze point position measuring device comprising a distortion correction means for correcting one point or a plurality of eye mark table values using values of a feature point table.