JP7324605B2 - 3D image generation device and its program - Google Patents

Description

Application of Article 30, Paragraph 2 of the Patent Law (1) Announced in the publication "2018 Institute of Image Information and Television Engineers Winter Conference Lecture Proceedings" published on December 6, 2018 by the Institute of Image Information and Television Engineers ( 2) Announced on December 21, 2018 at the 2018 ITE Winter Conference held by the Institute of Image Information and Television Engineers (ITE) from December 20 to 21, 2018.

The present invention relates to a stereoscopic image generating device and its program.

Integral photography (IP) methods have been studied since they do not require special spectacles and enable stereoscopic viewing with the naked eye. As shown in FIG. 7, a conventional IP stereoscopic image display device 100 includes a direct-view display 110 such as liquid crystal or organic EL, and a lens array 120 placed in front of the display. The lens array 120 is composed of element lenses 121 arranged two-dimensionally. The direct-view display 110 displays an elemental image group E composed of elemental images e corresponding to the elemental lenses 121 . Then, the elemental images e are projected in space by the elemental lenses 121 to form a stereoscopic image.

In the IP system, a very large number of pixels (amount of information) are required in order to reproduce images from many viewpoints. There are three parameters that determine the quality of IP stereoscopic video: the number of stereoscopic video pixels (the number of element images), the viewing zone, and the spatial frequency characteristics (depth reproduction range). As described in Non-Patent Document 1, it is known that these three parameters have a trade-off relationship. In other words, by narrowing the "viewing zone", it is possible to improve the "stereoscopic image pixel count" and the "spatial frequency characteristics". On the other hand, when the "viewing zone" is narrowed, the display range of stereoscopic images is narrowed. Also, in order to narrow the "visual zone", a lens array with a long focal length may be used.

H. Hoshino, F. Okano, and I. Yuyama, “Analysis of resolution limitation of integral photography,” J. Opt. Soc. Am. A 15(8), 2059-2065 (1998)

In the conventional IP method, if the direction of the "viewing zone" is controlled by a viewpoint tracking technique, which will be described later, stereoscopic images can be displayed in a wide range even if a lens array with a long focal length is used. In this case, as shown in FIG. 8, by translating the display position of the elemental image group E in the direction opposite to the viewer H, direction control of the "viewing zone" can be realized (reduction/enlargement is also possible). In FIG. 8A, since the viewer H has moved upward, the display position of the elemental image group E is moved downward, thereby moving the viewing zone shown by dots upward. Also, in FIG. 8B, since the viewer H has moved downward, the display position of the elemental image group E is moved upward, thereby moving the viewing zone shown by dots downward. In this way, in the IP system, it is possible to improve the "stereoscopic image pixel count" and "spatial frequency characteristics" and widen the "viewing zone".

In this viewpoint tracking technology, as shown in FIG. 9, a virtual display 210, a virtual lens array 220, a virtual camera array 230 consisting of a plurality of virtual cameras 231, and a three-dimensional model Obj as a subject are provided in a virtual space. to place. First, in the viewpoint tracking technique, pupil detection is performed, and the virtual camera array 230 is set so that the center of the visual field is located between the two eyes. That is, the central position of the virtual camera array 230 becomes the central position of the viewing area. In the viewpoint tracking technique, the virtual camera array 230 captures the multi-viewpoint image group M of the three-dimensional model Obj in various directions, and converts the captured multi-viewpoint image group M into the element image group E. This multi-viewpoint image group M is composed of images of different viewpoints (multi-viewpoint images m).

In FIG. 9, the white 'x' indicates the center position of the virtual camera array 230, and the black 'x' indicates the center position of both eyes detected by pupil detection. In FIG. 9 , since the center position of the virtual camera array 230 and the center position of both eyes detected by the pupil match, the white 'x' and the black 'x' are illustrated so as to overlap each other.

Here, in the viewpoint tracking technique, pupil detection may fail or rendering of the elemental image group E may be delayed depending on the timing at which the viewer H moves. With reference to FIG. 10, problems when they occur will be described. In FIG. 10, the actual positions of both eyes are indicated by symbols R and L. As shown in FIG. In addition, in FIG. 10, similarly to FIG. 9, white 'x' and black 'x' are illustrated so as to overlap each other.

As shown in FIG. 10, at time t=0, the viewer H is not moving, so the central position of the viewing area α (the central position of the virtual camera array 230) is positioned between the right eye R and the left eye L. Match. At time t=1, both the right eye R and the left eye L are within the viewing zone α. However, in the eye movement direction, the right eye R is approaching the boundary of the viewing zone α, and the margin β of the viewing zone α is decreasing. At time t=2, the viewing zone α cannot follow, and the right eye R deviates from the viewing zone α. As a result, crosstalk may occur and visibility may deteriorate. After that, at time t=3, since the movement of the viewer H has decreased, both the right eye R and the left eye L are within the viewing zone α.

As described above, in the viewpoint tracking technology, the directional control of the viewing zone cannot catch up with the movement of the viewer H, and the pupil may be out of the viewing zone of the stereoscopic image. As a result, crosstalk may occur and visibility may deteriorate.

Accordingly, it is an object of the present invention to provide a stereoscopic image generating apparatus and a program therefor in which the pupil is less likely to deviate from the viewing zone of the stereoscopic image.

SUMMARY OF THE INVENTION In view of the above-described problems, a stereoscopic image generating apparatus according to the present invention is a stereoscopic image generating apparatus that generates a stereoscopic image in which a viewing zone follows a person's pupil position from a group of multi-viewpoint images captured by a virtual camera array. The configuration includes virtual camera array setting means, pupil position prediction means, virtual camera array movement means, multi-viewpoint image group photographing means, and stereoscopic image generation means.

According to such a stereoscopic image generating apparatus, the virtual camera array setting means receives the pupil position information indicating the pupil position, and based on the inputted pupil position information, sets the center position of the virtual camera array to the position of the person's eyes. Set to middle position. That is, the virtual camera array setting means sets the central position of the virtual camera array so that the central position of the viewing zone is the middle position between the eyes of the person. Note that the center position of the virtual camera array is the center position of the viewing area.

The pupil position prediction means predicts the movement amount of the pupil position from the pupil position information.
The virtual camera array moving means moves the center position of the virtual camera array based on the movement amount of the pupil position predicted by the pupil position predicting means. That is, the virtual camera array moving means moves the center position of the virtual camera array considering the movement of the person (pupil).
The multi-viewpoint image group shooting means shoots the multi-viewpoint image group with the virtual camera array moved by the virtual camera array moving means.
The stereoscopic image generating means generates a stereoscopic image from the multi-viewpoint image group.
The pupil position predicting means calculates the moving speed of the pupil position from the time change of the pupil position indicated by the pupil position information, and calculates the amount of movement of the pupil position by multiplying the moving speed of the pupil position by a preset coefficient. .

The present invention can also be implemented by a program that causes hardware resources such as a CPU, memory, and hard disk provided in a computer to operate cooperatively as the stereoscopic image generation device described above.

According to the present invention, since the center position of the virtual camera array is moved in consideration of the movement of the person, the viewing area of the stereoscopic image follows the pupil position of the person, and the pupil is less likely to deviate from the viewing area of the stereoscopic image. As a result, it is possible to suppress the occurrence of crosstalk and the deterioration of visibility.

1 is an explanatory diagram of a stereoscopic image display system according to an embodiment; FIG. 1 is a schematic configuration diagram of a stereoscopic image display system according to an embodiment; FIG. 1 is a block diagram showing the configuration of a stereoscopic image generating device according to an embodiment; FIG. FIG. 4 is an explanatory diagram illustrating a method of generating an elemental image group in an embodiment; FIG. 10 is an explanatory diagram illustrating movement of a virtual camera array and tracking of a viewing zone in an embodiment; 4 is a flowchart showing the operation of the stereoscopic image generation device of FIG. 3; It is an explanatory view explaining a conventional IP stereoscopic video display device. FIG. 10 is an explanatory diagram for explaining a conventional viewpoint tracking technique; FIG. 10 is an explanatory diagram for explaining a conventional viewpoint tracking technique; FIG. 10 is an explanatory diagram for explaining setting of a virtual camera array in the prior art;

(embodiment)
[Overview of stereoscopic image display system]
An outline of a stereoscopic image display system 1 (FIG. 2) according to an embodiment of the present invention will be described with reference to FIGS.
As shown in FIG. 1, when controlling the direction of the viewing zone, the stereoscopic image display system 1 moves the center position of the virtual camera array 230 in consideration of the movement of the viewer (person). This makes it difficult for the eyes of the viewer to be out of the viewing zone.

In FIG. 1, the white 'x' indicates the center position of the virtual camera array 230, and the black 'x' indicates the center position of both eyes detected by pupil detection. In FIG. 1, since the center position of the virtual camera array 230 is moved from the center position of both eyes, the white 'x' and the black 'x' are shown at different positions.

As shown in FIG. 2 , the stereoscopic image display system 1 includes a pupil position detection device 2 , a stereoscopic image generation device 3 and a stereoscopic image display device 4 . In the present embodiment, it is assumed that the stereoscopic image display system 1 displays an IP stereoscopic image (element image group) and uses a known method for controlling the direction of the viewing zone itself (for example, see FIG. 8). .

The pupil position detection device 2 detects pupils of the viewer H and generates pupil position information indicating the detected pupil positions. For example, the pupil position detection device 2 includes a photographing device (for example, a web camera) for photographing the face image of the viewer H above the stereoscopic image display device 4 . Then, the pupil position detection device 2 detects the positions of the right eye R and the left eye L in real time from the face image captured by the imaging device using pupil detection technology (for example, OpenCV).
The pupil position detection device 2 outputs the generated pupil position information to the stereoscopic image generation device 3 . Here, when the pupil detection fails, the pupil position detection device 2 outputs pupil position information indicating the previously detected pupil position together with a notification to the effect that the pupil detection has failed.

The stereoscopic image generating device 3 generates an elemental image group E in which the viewing area is made to follow the pupil position of the viewer H from the multi-viewpoint image group M captured by the virtual camera array 230 . The stereoscopic image generation device 3 then outputs the generated elemental image group E to the stereoscopic image display device 4 . Details of the stereoscopic image generation device 3 will be described later.

The stereoscopic image display device 4 is a general IP stereoscopic display that displays the elemental image group E generated by the stereoscopic image generation device 3 . For example, the stereoscopic image display device 4 has a lens array 41 arranged in front of a direct-view display (not shown) such as LCD or OLED, as in the prior art. Here, in the stereoscopic image display system 1, since the visual field follows the pupil position of the viewer H, the quality of the elemental image group E can be improved even if the lens array 41 having a long focal length is used.

[Configuration of Stereoscopic Image Generation Device]
The configuration of the stereoscopic image generation device 3 will be described with reference to FIG.
As shown in FIG. 3, the stereoscopic image generation device 3 includes virtual camera array setting means 30, pupil movement amount calculation means (pupil position prediction means) 31, virtual camera array movement means 32, and three-dimensional model setting means 33. , multi-viewpoint image group photographing means 34 , and elemental image group generating means (stereoscopic image generating means) 35 .

The virtual camera array setting means 30 sets parameters regarding the virtual camera array 230 . For example, the parameters include the center position of the virtual camera array 230, the number of virtual cameras 231, their arrangement, their spacing, and their focal lengths. Specifically, the virtual camera array setting means 30 sets the center position of the virtual camera array 230 to the intermediate position between the viewer's H eyes based on the pupil position information input from the pupil position detection device 2. . In addition, the creator of the stereoscopic image may set the number, arrangement, spacing, and focal length of the virtual cameras 231 in the virtual camera array setting means 30 using operation means such as a mouse and keyboard (not shown). .
The virtual camera array setting means 30 outputs the set parameters to the virtual camera array moving means 32 .

The pupil movement amount calculator 31 predicts the movement amount of the pupil position from the pupil position information input from the pupil position detection device 2 . In this embodiment, the pupil movement amount calculation means 31 calculates the movement speed _vt of the pupil position from the time change of the pupil position indicated by the pupil position information. This moving speed v _t of the pupil position is obtained by dividing the change amount x _t −x _t′ of the pupil position by the time difference tt′, as shown in the following equation (1). Here, the detection times of the pupil positions are t and t', respectively, and the pupil positions at the detection times t and t' are _xt and _xt' (where t>t').

Then, the pupil movement amount calculation means 31 calculates the movement amount _kvt of the pupil position by multiplying the moving speed _vt of the pupil position by the coefficient k. This coefficient k is set in advance as an arbitrary value in consideration of, for example, the width of the viewing zone, the average movement speed of the pupil, and the success rate of pupil detection.
The pupil movement amount calculation means 31 outputs the movement amount _kvt of the pupil position to the virtual camera array movement means 32 .

The virtual camera array moving means 32 moves the center position of the virtual camera array 230 based on the movement amount _kvt of the pupil position input from the pupil movement amount calculating means 31 . That is, the virtual camera array moving means 32 moves the center position of the virtual camera array 230 from the center position of both eyes in consideration of the movement of the viewer H (pupils). The details of the virtual camera array moving means 32 will be described later (visual field tracking by moving the virtual camera array 230).
The virtual camera array moving means 32 outputs the center position of the moved virtual camera array 230 to the multi-viewpoint image group photographing means 34 together with the parameters input from the virtual camera array setting means 30 .

If the pupil position detection device 2 fails in pupil detection, the virtual camera array setting means 30 and the pupil movement amount calculation means 31 are notified that the pupil detection has failed and immediately preceding pupil position information is input. It is assumed that This pupil position information indicates the previous pupil position detected by the pupil position detection device 2 .
Further, when it is notified that the pupil detection has failed, the pupil movement amount calculation means 31 does not calculate the movement amount _kvt of the pupil position, and the virtual camera array movement means 32 does not move the center position of the virtual camera array 230. It is assumed that

The three-dimensional model setting means 33 sets a three-dimensional model Obj to be arranged in the virtual space. This three-dimensional model Obj is an object to be photographed by the virtual camera array 230, and is arbitrary like a woman wearing a hat (FIG. 1). For example, the creator of the stereoscopic image sets the three-dimensional model Obj to be arranged in the virtual space in the three-dimensional model setting means 33 using the operation means.
The three-dimensional model setting means 33 outputs the set three-dimensional model Obj to the multi-viewpoint image group photographing means 34 .

The multi-viewpoint image group shooting means 34 shoots the multi-viewpoint image group M with the virtual camera array 230 moved by the virtual camera array moving means 32 . Specifically, the multi-viewpoint image group photographing means 34 photographs the virtual space in which the three-dimensional model Obj is arranged, with the virtual camera array 230 that is moved by the movement amount _kvt of the pupil position from the center position of both eyes. In this manner, the multi-viewpoint image group photographing means 34 generates (renders) the multi-viewpoint image group M. FIG.
The multi-viewpoint image group photographing means 34 outputs the generated multi-viewpoint image group to the element image group generation means 35 .

The elemental image group generating means 35 generates an elemental image group E from the multi-viewpoint image group M input from the multi-viewpoint image group photographing means 34 . The elemental image group generating means 35 can generate the elemental image group E from the multi-viewpoint image group M by any method. For example, the elemental image group generating means 35 extracts pixels at the same position from each multi-viewpoint image m, and generates an elemental image e composed of the extracted pixels.
The element image group generating means 35 outputs the generated element image group E to the stereoscopic image display device 4 .

<Generation of elemental image group>
An example of generation of the elemental image group E will be described with reference to FIG.
As shown in FIG. 4A, for a multi-viewpoint image m for three viewpoints, namely, an image m ₁₁ at the upper left viewpoint, an image m ₂₁ at the right viewpoint of the image m ₁₁ , and an image m ₁₂ at the lower viewpoint of the image m ₁₁ . think. In order to simplify the explanation, in the image _m11 , the upper left pixel is marked with "● 1", the right pixel is marked with "● 2", and the lower pixel is marked with "● 3". Similarly, " _▴1 " to "▴3" are added to the image m21, and "▪1" to "▪3" are added to the image _m12 .

In this case, as shown in FIG _. 4(b), the elemental image group generating means 35 extracts upper left pixels ●1, ▲ ₁ , _■ 1, . Generate image _e11 . That is, the element image _e11 is composed of upper left pixels ●1, ▲1, ■1, . . . at the same position. Also, in the element image _e11 , the arrangement of the upper left pixels ● ₁ , ▲ ₁ , ■ ₁ , . In other words, in the element image _e11 , the pixel ●1 is arranged on the upper left, the pixel ▲1 is arranged on the right side of the pixel ●1, and the pixel ■1 is arranged below the pixel ●1.

Further, the element image group generating means ₃₅ generates an element image _e21 composed of right pixels 2, _{2 ,} 2 _, . . . Generate. Furthermore, the element image generating means ₃₆ generates element images composed _of lower pixels ● ₃ , ▲ ₃ , ■ ₃ , . . . e ₁₂ is generated.

Here, the elemental image e is arranged at a position corresponding to the pixel position of each pixel forming the elemental image e. For example, the element image _e11 is composed of the upper left pixels ●1, ▲1, ■1, . . . Also, since the element image _e21 is composed of the right pixels ●2, ▲2, ■ ₂ , . Further, the element image _e12 is arranged below the element image _e11 because it is composed of the lower pixels ●3, ▲3, ■3, .

Depending on the configuration of the optical system of the IP stereoscopic display device, an inverted stereoscopic image may be displayed. In this case, the elemental image group generating means 35 may generate the elemental image group E so that a stereoscopic erect image is displayed by inverting the position of each pixel vertically and horizontally.

<Viewing area tracking by moving the virtual camera array>
With reference to FIG. 5, viewing zone tracking by movement of the virtual camera array will be described.
In FIG. 5, the actual positions of both eyes are indicated by symbols R and L. As shown in FIG. In FIG. 5, similarly to FIG. 1, a white 'x' indicating the central position of the virtual camera array 230 and a black 'x' indicating the central position of both eyes detected by the pupil are illustrated. Note that the center position of the virtual camera array 230 (white 'x') is the center position of the viewing area α.

As shown in FIG. 5, at time t=0, the viewer H is not moving, so the central position of the viewing zone α coincides with the intermediate position between the right eye R and the left eye L. As shown in FIG. Therefore, at time t=0, the white 'x' and the black 'x' overlap.
At time t=1, both the right eye R and the left eye L are within the viewing zone α. At this time, since the virtual camera array 230 is moved by the movement amount _kv1 , the right eye R is sufficiently separated from the boundary of the viewing area α, and the margin β of the viewing area α has more margin than in FIG. be.

At time t=2, pupil detection fails, so the viewing zone α has not moved from time t=1, but the pupil of the right eye R is within the viewing zone α. As a result, the stereoscopic image generation device 3 can suppress the occurrence of crosstalk and deterioration of visibility.
At time t=3, both the right eye R and the left eye L are within the viewing zone α. Also at this time, the virtual camera array 230 is moved by the movement amount _kv3 .
When the viewer H moves to the opposite side (leftward in FIG. 5), the sign of the movement amount _kvt should be negative.

[Operation of Stereoscopic Image Generation Device]
The operation of the stereoscopic image generation device 3 will be described with reference to FIG.
As shown in FIG. 6, in step S1, the virtual camera array setting means 30 sets the central position of the virtual camera array 230 to the viewer H's both eyes based on the pupil position information input from the pupil position detection device 2. set to the middle position of
In step S<b>2 , the pupil movement amount calculator 31 predicts the movement amount of the pupil position from the pupil position information input from the pupil position detection device 2 . For example, the pupil movement amount calculation means 31 calculates the movement speed of the pupil position from the change in the pupil position indicated by the pupil position information, as shown in the above equation (1), and calculates the calculated movement speed of the pupil position and the coefficient. By multiplying, the movement amount of the pupil position is calculated.

In step S3, the virtual camera array moving means 32 moves the center position of the virtual camera array based on the movement amount of the pupil position calculated in step S2. That is, the virtual camera array moving means 32 moves the center position of the virtual camera array from the center position of both eyes in consideration of the viewer's H movement.
At step S4, the three-dimensional model setting means 33 sets a three-dimensional model to be arranged in the virtual space.

In step S5, the multi-viewpoint image group shooting means 34 shoots a multi-viewpoint image group with the virtual camera array moved in step S3. Specifically, the multi-viewpoint image group photographing means 34 photographs the virtual space in which the three-dimensional model is arranged, with a virtual camera array moved from the center position of both eyes by the movement amount of the pupil position.
In step S6, the elemental image group generating means 35 generates an elemental image group from the multi-viewpoint image group photographed in step S5. For example, the elemental image group generating means 35 extracts pixels at the same position from each multi-viewpoint image, and generates elemental images composed of the extracted pixels.

[Action/effect]
As described above, the stereoscopic image generation device 3 according to the embodiment of the present invention moves the center position of the virtual camera array in consideration of the movement of the viewer H. area follows, and the pupil is less likely to deviate from the viewing area of the stereoscopic image. As a result, the stereoscopic image generation device 3 can suppress the occurrence of crosstalk and deterioration of visibility, and can provide a high-quality IP stereoscopic image.

(Modification: Movement amount of pupil position)
Although the embodiments of the present invention have been described in detail above, the present invention is not limited to the above-described embodiments, and includes design changes and the like without departing from the gist of the present invention.
In the above-described embodiment, the pupil movement amount calculation means calculates the movement amount of the pupil position using Equation (1), but the present invention is not limited to this. For example, the pupil movement amount calculation means may calculate the movement amount of the pupil position using a higher-order expression regarding the moving speed v of the pupil position, as shown in the following equation (2). Note that the coefficient l is set in advance to an arbitrary value, like the coefficient k.

Further, the pupil movement amount calculation means may use something other than the movement speed, such as the acceleration of the pupil position, as long as it relates to the movement of the pupil position.
Further, in the above equation (1), the movement speed of the pupil position is calculated at intervals of one frame, but the movement speed of the pupil position may be calculated at intervals of two frames or more.

(Other modifications)
In the above-described embodiment, the present invention is applied to the IP stereoscopic method having omnidirectional parallax, but the present invention is not limited to this. INDUSTRIAL APPLICABILITY The present invention can be applied to a light-reproducing stereoscopic system such as an IP stereoscopic system and a lenticular stereoscopic system. This lenticular stereoscopic system is a light-reproducing stereoscopic system having only horizontal parallax.

In the above-described embodiment, the pupil positions of one person are detected, but the present invention is not limited to this. In other words, the pupil positions of a plurality of people are detected, and time-division display is performed according to the number of people whose pupil positions are detected. In other words, the stereoscopic image generation device may include a time division unit that time-divides the stereoscopic image generated by the stereoscopic image generation unit for each person.

In each of the above-described embodiments, the stereoscopic image generation device has been described as independent hardware, but the present invention is not limited to this. For example, the present invention can be realized by a program that causes hardware resources such as a CPU, memory, and hard disk provided in a computer to operate cooperatively as the stereoscopic image generation device described above. These programs may be distributed via a communication line, or may be distributed after being written in a recording medium such as a CD-ROM or flash memory.

1 stereoscopic image display system 2 pupil position detection device 3 stereoscopic image generation device 4 stereoscopic image display device 30 virtual camera array setting means 31 pupil movement amount calculation means (pupil position prediction means)
32 virtual camera array moving means 33 three-dimensional model setting means 34 multi-viewpoint image group photographing means 35 elemental image group generating means (stereoscopic image generating means)
41 lens array 100 IP stereoscopic video display device 110 direct-view display 120 lens array 121 element lens 210 virtual display 220 virtual lens array 230 virtual camera array 231 virtual camera E element image group e element image L left eye M multi-viewpoint image group m multiple Viewpoint image H viewer Obj 3D model R right eye α viewing zone

Claims (4)

A stereoscopic image generation device for generating a stereoscopic image in which a viewing zone is made to follow a person's pupil position from a group of multi-viewpoint images captured by a virtual camera array,
virtual camera array setting means for receiving pupil position information indicating the pupil position, and setting the center position of the virtual camera array to an intermediate position between the eyes of the person based on the inputted pupil position information;
Pupil position prediction means for predicting a movement amount of the pupil position from the pupil position information;
virtual camera array movement means for moving the center position of the virtual camera array based on the movement amount of the pupil position predicted by the pupil position prediction means;
multi-viewpoint image group photographing means for photographing the multi-viewpoint image group with the virtual camera array moved by the virtual camera array moving means;
stereoscopic image generation means for generating the stereoscopic image from the multi-viewpoint image group;
with
The pupil position prediction means calculates the moving speed of the pupil position from the temporal change of the pupil position indicated by the pupil position information, and multiplies the moving speed of the pupil position by a preset coefficient to obtain the pupil position. A stereoscopic image generating device, characterized by calculating a movement amount . A stereoscopic image generation device for generating a stereoscopic image in which a viewing zone is made to follow a person's pupil position from a group of multi-viewpoint images captured by a virtual camera array,
virtual camera array setting means for receiving pupil position information indicating the pupil position, and setting the center position of the virtual camera array to an intermediate position between the eyes of the person based on the inputted pupil position information;
pupil position prediction means for predicting a movement amount of the pupil position from the pupil position information;
virtual camera array movement means for moving the center position of the virtual camera array based on the movement amount of the pupil position predicted by the pupil position prediction means;
multi-viewpoint image group photographing means for photographing the multi-viewpoint image group with the virtual camera array moved by the virtual camera array moving means;
stereoscopic image generation means for generating the stereoscopic image from the multi-viewpoint image group;
a time-division means for time-dividing the stereoscopic image generated by the stereoscopic image generation means for each person ;
A stereoscopic image generating device comprising: 3. The stereoscopic image generating apparatus according to claim 1, wherein the stereoscopic image is a group of element images corresponding to a light-reproduction stereoscopic system. A program for causing a computer to function as the stereoscopic image generating device according to any one of claims 1 to 3 .

Images