Line-of-sight tracking system and method
A display system includes a line-of-sight tracking camera to detect an eyeball of a driver, first and second stereo cameras, a controlling unit, and a storing unit. The first and second stereo camera photographs a range corresponding to a field-of-view based on stereo camera-based line-of-sight information that is changed based on line-of-sight tracking camera-based line-of-sight information, and provide a photographed image. The controlling unit converts the line-of-sight tracking camera-based line-of-sight information into first and second stereo camera-based line-of-sight information based on pre-stored position and rotation information of the tracking camera and the first and second stereo cameras, and project the converted information onto the first and second stereo cameras, such that a three-dimensional line-of-sight coordinate is calculated. The storing unit stores information related to a system as well as the position and rotation information of the tracking camera and the first and second stereo cameras.
1. A display system, comprising:
a line-of-sight tracking camera configured to detect an eyeball of a driver;
first and second stereo cameras configured to photograph a range corresponding to a driver's field-of-view based on stereo camera-based driver's line-of-sight information that is changed based on line-of-sight tracking camera-based driver's line-of-sight information, and provide a photographed image;
a controlling unit configured to convert the line-of-sight tracking camera-based driver's line-of-sight information into first and second stereo camera-based driver's line-of-sight information based on pre-stored position information and rotation information of the line-of-sight tracking camera and the first and second stereo cameras, and project the converted information onto the first and second stereo cameras, such that a three-dimensional coordinate of a driver's line-of-sight is calculated; and
a storing unit configured to store information related to the display system as well as the position information and the rotation information of the line-of-sight tracking camera and the first and second stereo cameras.
2. The display system according to claim 1 , wherein:
the line-of-sight tracking camera-based driver's line-of-sight information includes a line-of-sight tracking vector monocular eye point and a line-of-sight tracking vector, and
the controlling unit is configured to convert the line-of-sight tracking camera-based line-of-sight tracking vector monocular eye point into a stereo camera-based line-of-sight tracking vector monocular eye point.
3. The display system according to claim 2 , wherein the controlling unit is configured to convert the line-of-sight tracking camera-based line-of-sight tracking vector monocular eye point (Location) (x loc eye , y loc eye , z loc eye , 1) into a stereo camera-based line-of-sight tracking vector monocular eye point (x loc global , y loc global , z loc global , 1) through the following Equation 1:
Location
=
[
1
0
0
t
xyz
x
0
1
0
t
xyz
y
0
0
1
t
xyz
z
0
0
0
1
]
︸
Position
Movement
[
1
0
0
0
0
cos
θ
x
-
sin
θ
x
0
0
sin
θ
x
cos
θ
x
0
0
0
0
1
]
︸
X
axis
rotation
[
cos
θ
y
0
-
sin
θ
y
0
0
1
0
0
sin
θ
y
0
cos
θ
y
0
0
0
0
1
]
︸
Y
axis
rotation
[
cos
θ
z
-
sin
θ
z
0
0
sin
θ
z
cos
θ
z
0
0
0
0
1
0
0
0
0
1
]
︸
Z
axis
rotation
[
x
loc
eye
y
loc
eye
z
loc
eye
1
]
where t xyz x , t xyz y , and t xyz z mean position information, and Θ x , Θ y , Θ z mean rotation information.
4. The display system according to claim 1 , wherein:
the line-of-sight tracking camera-based driver's line-of-sight information includes a line-of-sight tracking vector monocular eye point and a line-of-sight tracking vector, and
the controlling unit is configured to convert the line-of-sight tracking camera-based line-of-sight tracking vector into a stereo camera-based line-of-sight tracking vector.
5. The display system according to claim 4 , wherein the controlling unit is configured to convert the line-of-sight tracking camera-based line-of-sight tracking vector (x vec eye ,y vec eye ,z vec eye ,1) into a stereo camera-based line-of-sight tracking vector (Rotation) (x vec global ,y vec global ,z vec global ,1) through the following Equation 2:
Rotation
=
[
1
0
0
0
0
cos
θ
x
-
sin
θ
x
0
0
sin
θ
x
cos
θ
x
0
0
0
0
1
]
︸
X
axis
rotation
[
cos
θ
y
0
-
sin
θ
y
0
0
1
0
0
sin
θ
y
0
cos
θ
y
0
0
0
0
1
]
︸
Y
axis
rotation
[
cos
θ
z
-
sin
θ
z
0
0
sin
θ
z
cos
θ
z
0
0
0
0
1
0
0
0
0
1
]
︸
Z
axis
rotation
[
x
vec
eye
y
vec
eye
z
vec
eye
1
]
where Θ x , Θ y , Θ z mean rotation information.
6. The display system according to claim 1 , wherein the controlling unit is configured to calculate a driver's line-of-sight gaze point from a stereo camera-based line-of-sight tracking vector projected onto the first and second stereo cameras and calculate the three-dimensional coordinate of the driver's line-of-sight based on the calculated driver's line-of-sight gaze point.
7. The display system according to claim 6 , wherein the controlling unit is configured to generate a window corresponding to a preset main gaze range of a person's line-of-sight on an image projected onto the first stereo camera, move a monocular eye point of the generated window according to a first stereo camera-based line-of-sight tracking vector, perform template matching such that the moved window corresponds to a line-of-sight tracking vector of the second stereo camera, and recognize a highest position as a driver's line-of-sight gaze point as a result of the performing the template matching.
8. The display system according to claim 1 , wherein the three-dimensional coordinate of the driver's line-of-sight is P(Xp, Yp, Zp),
X
p
=
T
d
,
Y
p
=
T
d
,
and
Z
p
=
f
T
-
,
where P 1 (x 1 , y 1 ) is a point at which P is projected onto an imaging surface of the first stereo camera, P r (x r , y r ) is a point at which P is projected onto an imaging surface of the second stereo camera, f is a focal length of the camera, T is a distance between the first and second stereo cameras, and d is a value obtained by dividing a distance of a coordinate measuring point by the focal length of the camera.
9. A display method providing a three-dimensional driver's line-of-sight in a display system including a line-of-sight tracking camera and first and second stereo cameras, the display method comprising:
detecting an eyeball of a driver through the line-of-sight tracking camera to recognize line-of-sight tracking camera-based driver's line-of-sight information;
converting the line-of-sight tracking camera-based driver's line-of-sight information into first and second stereo camera-based driver's line-of-sight information based on pre-stored position information and rotation information of the line-of-sight tracking camera and the first and second stereo cameras;
recognizing a driver's line-of-sight gaze point based on the first and second stereo camera-based driver's line-of-sight information; and
converting the driver's line-of-sight gaze point into a three-dimensional coordinate of a driver's line-of-sight.
10. The display method according to claim 9 , wherein:
the line-of-sight tracking camera-based driver's line-of-sight information includes a line-of-sight tracking vector monocular eye point and a line-of-sight tracking vector, and
the converting of the line-of-sight tracking camera-based driver's line-of-sight information into the first and second stereo camera-based driver's line-of-sight information includes converting the line-of-sight tracking camera-based line-of-sight tracking vector monocular eye point into a stereo camera-based line-of-sight tracking vector monocular eye point.
11. The display method according to claim 10 , wherein the converting of the line-of-sight tracking camera-based driver's line-of-sight information into the first and second stereo camera-based driver's line-of-sight information includes converting the line-of-sight tracking camera-based line-of-sight tracking vector monocular eye point (Location) (x loc eye , y loc eye , z loc eye ,1) into a stereo camera-based line-of-sight tracking vector monocular eye point (x loc global , y loc global , z loc global , 1) through the following Equation 1:
Location
=
[
1
0
0
t
xyz
x
0
1
0
t
xyz
y
0
0
1
t
xyz
z
0
0
0
1
]
︸
Position
Movement
[
1
0
0
0
0
cos
θ
x
-
sin
θ
x
0
0
sin
θ
x
cos
θ
x
0
0
0
0
1
]
︸
X
axis
rotation
[
cos
θ
y
0
-
sin
θ
y
0
0
1
0
0
sin
θ
y
0
cos
θ
y
0
0
0
0
1
]
︸
Y
axis
rotation
[
cos
θ
z
-
sin
θ
z
0
0
sin
θ
z
cos
θ
z
0
0
0
0
1
0
0
0
0
1
]
︸
Z
axis
rotation
[
x
loc
eye
y
loc
eye
z
loc
eye
1
]
where t xyz x , t xyz y , and t xyz z mean position information, and Θ x , Θ y , Θ z mean rotation information.
12. The display method according to claim 9 , wherein:
the line-of-sight tracking camera-based driver's line-of-sight information includes a line-of-sight tracking vector monocular eye point and a line-of-sight tracking vector, and
the converting of the line-of-sight tracking camera-based driver's line-of-sight information into the first and second stereo camera-based driver's line-of-sight information includes converting the line-of-sight tracking camera-based line-of-sight tracking vector into a stereo camera-based line-of-sight tracking vector.
13. The display method according to claim 12 , wherein the converting of the line-of-sight tracking camera-based driver's line-of-sight information into the first and second stereo camera-based driver's line-of-sight information includes converting the line-of-sight tracking camera-based line-of-sight tracking vector (x vec eye ,y vec eye ,z vec eye ,1) into a stereo camera-based line-of-sight tracking vector (Rotation) (x vec global ,y vec global ,z vec global ,1) through the following Equation 2:
Rotation
=
[
1
0
0
0
0
cos
θ
x
-
sin
θ
x
0
0
sin
θ
x
cos
θ
x
0
0
0
0
1
]
︸
X
axis
rotation
[
cos
θ
y
0
-
sin
θ
y
0
0
1
0
0
sin
θ
y
0
cos
θ
y
0
0
0
0
1
]
︸
Y
axis
rotation
[
cos
θ
z
-
sin
θ
z
0
0
sin
θ
z
cos
θ
z
0
0
0
0
1
0
0
0
0
1
]
︸
Z
axis
rotation
[
x
vec
eye
y
vec
eye
z
vec
eye
1
]
where Θ x , Θ y , Θ z mean rotation information.
14. The display method according to claim 9 , wherein the recognizing of the driver's line-of-sight gaze point includes:
projecting the first and second stereo camera-based driver's line-of-sight information onto the first and second stereo cameras; and
recognizing the driver's line-of-sight gaze point based on the information projected onto the first and second stereo cameras.
15. The display method according to claim 14 , wherein the recognizing of the driver's line-of-sight gaze point based on the information projected onto the first and second stereo cameras includes:
generating a window corresponding to a preset main gaze range of a person's line-of-sight on an image projected onto the first stereo camera;
moving a monocular eye point of the generated window according to a first stereo camera-based line-of-sight tracking vector;
performing template matching such that the moved window corresponds to a line-of-sight tracking vector of the second stereo camera; and
recognizing a highest position as a driver's line-of-sight gaze point as a result of the performance.
16. The display method according to claim 14 , wherein the three-dimensional coordinate of the driver's line-of-sight is P(Xp, Yp, Zp),
X
p
=
T
d
,
Y
p
=
T
d
,
and
Z
p
=
f
T
-
,
where P 1 (x 1 , y 1 ) is a point at which P is projected onto an imaging surface of the first stereo camera, P r (x r , y r ) is a point at which P is projected onto an imaging surface of the second stereo camera, f is a focal length of the camera, T is a distance between the first and second stereo cameras, and d is a value obtained by dividing a distance of a coordinate measuring point by the focal length of the camera.