JP7499078B2 - Stereoscopic image depth control device and its program - Google Patents

Description

The present invention relates to a stereoscopic image depth control device and its program that compresses the depth of a three-dimensional model of a subject.

An integral 3D display is a flat display that displays elemental images and has a lens array superimposed on it, making it possible to display an object as if it were actually there, without the need for special glasses. By reproducing light rays in three-dimensional space, the integral method can naturally present horizontal and vertical binocular parallax and motion parallax without glasses. In addition, because the integral method reproduces light rays emanating from an object, it can avoid the main problem of 3D displays, which is the mismatch between convergence and accommodation.

On the other hand, spatial image reproduction 3D displays such as the integral system have the problem that it is difficult to present a scene with a wide depth. Specifically, the spatial frequency decreases as the distance in the depth direction from the lens array surface increases, so when an image is displayed at a distance, the image becomes significantly blurred. As shown in FIG. 8, the resolution (spatial frequency γ) of a stereoscopic image reproduced by the integral system maintains a predetermined resolution (Nyquist frequency) in a constant depth reproduction range before and after the lens array position (z = 0), and has the characteristic of rapidly decreasing when the range is exceeded. This predetermined resolution is specified by the spacing (pitch) of the element lenses that make up the lens array, and a stereoscopic image that exceeds this depth reproduction range becomes blurred as it moves away from the position of the lens array. This means that in the integral system, in principle, there is a depth reproduction range in which a stereoscopic image can be displayed without blurring.

In order to expand the area (depth reproduction range) where no blurring occurs, such as images of the same subject, it is necessary to reduce the pixel pitch of the elemental images, which places a burden on system development (Non-Patent Document 1). Therefore, as shown in FIG. 9, a conventional method has been proposed to compress the depth of a 3D scene and avoid blurring (Non-Patent Document 2). In this conventional method, as shown in the following formula (1), the vertex coordinates p (x, y, z) constituting an object (three-dimensional model M) in the scene are converted to vertex coordinates p ₀ ' (x ₀ ', y ₀ ', z ₀ '). At that time, the image is moved from the back of the screen to the front as viewed from the viewer 9, and a scene with a wide depth is compressed to a narrow range that can be displayed without blurring.

In equation (1), f(z) is a function that moves a certain vertex coordinate from a depth position z on the far side to a depth position z' on the near side as viewed from the viewer 9, as shown in Fig. 10. Also, in equation (1), the transformation of _x0 ', _y0 ' is a transformation that reduces the object so that the apparent size of the object does not change before and after compression as viewed from the origin of depth compression.

Note that p _v0 (0,0,0) is the initial viewing position (standard viewing position). The horizontal and vertical positions of the point where the vertex coordinates are projected onto the display are (i,j), respectively. The subscript pre indicates the projection of the vertex before compression, and the subscript post indicates the projection of the vertex after compression. The subscript number 0 indicates the projection of the initial viewing position. The subscript number 1 indicates the projection of the viewing position after movement.

A method has been proposed to determine a nonlinear depth compression function by calculating the pixel spacing and lens element spacing of an integral 3D display, as well as the depth range (depth reproduction range) that can be displayed without blurring (Patent Document 1). Hereinafter, static compression of the depth of a scene without considering movement of the viewing position, as in this conventional technology, will be referred to as "static depth compression."

In this static depth compression, when the viewing position is moved, the shape distortion of the object becomes noticeable. As shown in FIG. 11, at the initial viewing position p _v0 (0,0,0), points p ₀ '(x ₀ ', y ₀ ', z ₀ ') and p ₁ '(x ₁ ', y ₁ ', z ₁ ') remain at the same positions before and after the depth compression. On the other hand, at the viewing position p _v1 (x _v1 , y _v1 , z _v1 ) after the movement, the projection points (i _pre1 , j _pre1 ) and (i _post1 , j _post1 ) on the display are shifted before and after the depth compression, and the shape of the object appears distorted. In this way, in the static depth compression, it is difficult to notice that an object with a compressed depth is compressed when viewed from the front, but the distortion of the shape due to the compression becomes noticeable when viewed from the side.

Here, the shape distortion is defined as the deviations D _shape0 and D shape1 of the points projected onto the integral 3D display at the initial viewing position and the viewing position after movement, respectively, as shown in the following formula ( ₂ ).

Thus, in static depth compression, no shape distortion occurs at the initial viewing position, and D _shape0 = 0. On the other hand, in static depth compression, when the viewing position moves, D _shape1 > 0, and shape distortion increases.

Therefore, as shown in Fig. 12, a method has been proposed to dynamically compress the depth of a scene by taking into account the movement of the viewing position, which will be referred to as "dynamic depth compression" hereafter (Non-Patent Document 3). In this dynamic depth compression, in order to measure the viewing position in real time, an image of the viewer 9 is captured by an infrared camera or an RGB camera, and the image is processed. In addition, in dynamic depth compression, the vertex coordinates p(x, y, z) of an object are converted to _p1 '( _x1 ', _y1 ', _z1 ') based on the following formula (3).

In this dynamic depth compression, as shown in Fig. 12, the projection points (i _pre1 , j _pre1 ), (i _post1 , j _post1 ) always match before and after the depth compression, so no distortion of the shape occurs. Similarly, the projection points (i _pre0 , j _pre0 ), (i _post0 , j _post0 ) also match before and after the depth compression. Therefore, D _shape0 =0, D _shape1 =0.

However, with this dynamic depth compression, when the viewing position is moved, an object that should actually appear stationary in the air may appear to slide in the same direction as the movement of the viewing position. It is known that shear distortion caused by the movement of the viewing position occurs in a binocular stereoscopic display when the distance between the shooting cameras does not match the distance between the eyes of the viewer 9 (Non-Patent Document 4). Here, spatial distortion is defined as shown in the following equation (4).

There is a trade-off between the spatial distortion and the shape distortion described above. For example, in static depth compression, D _shear1 =0 and D _shape1 >0, whereas in dynamic depth compression, D _shear1 >0 and D _shape1 =0.

JP 2017-11520 A

Hoshino H, Okano F, Isono H, Yuyama , Analysis of resolution limitation of integral photography, Opt Soc Am, 1998,15(8),2059-2065 Sawahata Y, Morita , Estimating Depth Range Required for 3-D Displays to Show Depth-Compressed Scenes Without Inducing Sense of Unnaturalness, IEEE Trans Broadcast, 2018,64(2),488-497 Miyashita Y, Sawahata Y, Katayama M, Komine , Depth boost: Extended depth reconstruction capability on volumetric display, In: ACM SIGGRAPH 2019 Talks, SIGGRAPH 2019,2019 Wartell Z, Hodges LF, Ribarsky , Balancing fusion, image depth and distortion in stereoscopic head-tracked displays, Proc 26th Annu Conf Comput Graph Interact Tech SIGGRAPH 1999, 1999,351-358

However, the conventional technology described above does not have a mechanism for adjusting spatial distortion and geometric distortion in depth compression representation of 3D images, and only one of the distortions can be selected at its maximum.

Therefore, the objective of the present invention is to provide a stereoscopic image depth control device and associated program that can adjust the ratio of spatial distortion to geometric distortion.

To solve the above problem, the stereoscopic image depth control device of the present invention compresses the depth of a three-dimensional model of a subject so that it falls within a depth reproduction range that satisfies a preset resolution, and generates element images from the depth-compressed three-dimensional model, and is configured to include a static vertex coordinate calculation unit, a dynamic vertex coordinate calculation unit, a vertex coordinate adjustment unit, and an element image generation unit.

According to this configuration, the static vertex coordinate calculation section calculates static vertex coordinates, which are the coordinates of each vertex constituting a three-dimensional model, by static depth compression that does not take into account movement of the viewing position of the stereoscopic image.
The dynamic vertex coordinate calculation unit calculates dynamic vertex coordinates, which are the coordinates of each vertex constituting a three-dimensional model according to the viewing position, by dynamic depth compression that takes into account the movement of the viewing position.
The vertex coordinate adjustment unit adjusts the vertex coordinates of the three-dimensional model between the static vertex coordinates and the dynamic vertex coordinates based on an adjustment ratio between the static depth compression and the dynamic depth compression.
The element image generating unit generates element images from the three-dimensional model after the vertex coordinates have been adjusted.
Static depth compression that does not take into account movement of the viewing position is expressed by equation (1) which converts the vertex coordinates p( x, y, z) of a three-dimensional model into vertex coordinates p ₀ '(x ₀ ', y ₀ ', z ₀ ').
Dynamic depth compression that takes into account movement of the viewing position is expressed by equation (3) that converts the vertex coordinates p(x, y, z) of a three-dimensional model into vertex coordinates p ₁ '(x ₁ ', y ₁ ', z ₁ ').

In other words, the adjustment ratio represents the ratio between the shape distortion caused by static depth compression and the space distortion caused by dynamic depth compression. Therefore, the stereoscopic image depth control device can adjust the ratio between the space distortion and the shape distortion in the vertex coordinate adjustment unit so that both can be mitigated.

The present invention can also be realized by a program that causes a computer to function as the stereoscopic image depth control device described above.

The present invention allows you to adjust the ratio of spatial distortion to geometric distortion.

1 is a schematic configuration diagram of a stereoscopic image display system according to a first embodiment. FIG. 2 is an explanatory diagram illustrating an example of an adjustment ratio input device in the first embodiment. 1 is a block diagram showing a configuration of a stereoscopic image depth control device according to a first embodiment. 4 is an explanatory diagram for explaining adjustment of vertex coordinates of a three-dimensional model in the first embodiment. FIG. 5 is an explanatory diagram illustrating the relationship between the setting of an adjustment ratio and distortion of shape and distortion of space in the first embodiment. FIG. 4 is a flowchart showing the operation of the stereoscopic image depth control device according to the first embodiment. FIG. 11 is a block diagram showing the configuration of a stereoscopic image depth control device according to a second embodiment. FIG. 1 is an explanatory diagram illustrating a depth reproduction range in a conventional integral 3D display. FIG. 1 is an explanatory diagram illustrating conventional static depth compression. 1 is a graph showing an example of a function for compressing depth in conventional static depth compression. FIG. 1 is an explanatory diagram illustrating shape distortion in conventional static depth compression. FIG. 1 is an explanatory diagram illustrating conventional dynamic depth compression.

Each embodiment of the present invention will be described below with reference to the drawings. However, each embodiment described below is intended to embody the technical concept of the present invention, and unless otherwise specified, the present invention is not limited to the following. In addition, in each embodiment, the same means are given the same reference numerals, and the description may be omitted.

First Embodiment
[Configuration of stereoscopic image display system]
The configuration of a stereoscopic image display system 100 according to the first embodiment will be described with reference to FIG.
The stereoscopic image display system 100 generates elemental images from a three-dimensional model of a subject and displays the generated elemental images. At this time, the stereoscopic image display system 100 compresses the depth of the three-dimensional model in a stereoscopic image depth control device 4 described later so that the depth fits within the depth reproduction range of the stereoscopic image display device 1. As shown in FIG. 1, the stereoscopic image display system 100 includes a stereoscopic image display device 1, a viewing position measurement device 2, an adjustment ratio input device 3, and a stereoscopic image depth control device 4.

The stereoscopic image display device 1 is a typical integral 3D display that displays a stereoscopic image by the integral method. This stereoscopic image display device 1 inputs elemental images from a stereoscopic image depth control device 4 and displays the elemental images via a lens array (not shown), thereby allowing the viewer 9 to view a stereoscopic image T.

The viewing position measuring device 2 measures the position (viewing position) at which the viewer 9 views the stereoscopic image T. This viewing position measuring device 2 can be composed of an infrared camera or an RGB camera. Specifically, the viewing position measuring device 2 may detect the left and right corneas of the viewer 9 from camera images of the viewer 9 taken with two RGB cameras using the scleral reflex method or the corneal-pupillary reflex method, determine the three-dimensional viewing position using the principle of triangulation, and measure the viewing distance from the stereoscopic image display device 1 to the viewer 9. The viewing position measuring device 2 then inputs the measured viewing position and viewing distance to the stereoscopic image depth control device 4.

The adjustment ratio input device 3 allows the viewer 9 to set an adjustment ratio s, which is then input to the stereoscopic image depth control device 4 (vertex coordinate adjustment unit 43). For example, as shown in FIG. 2, the adjustment ratio input device 3 is a slider that can set the adjustment ratio s to any value between 0 and 1 (0≦s≦1). This adjustment ratio s represents the ratio between static depth compression and dynamic depth compression. In other words, the adjustment ratio s represents the ratio between the shape distortion caused by static depth compression and the spatial distortion caused by dynamic depth compression.

The stereoscopic image depth control device 4 compresses the depth of the three-dimensional model of the subject so that it falls within a depth reproduction range that satisfies a preset resolution of the stereoscopic image display device 1, and generates elemental images from the depth-compressed three-dimensional model. At this time, the stereoscopic image depth control device 4 adjusts the ratios of both the spatial distortion and the shape distortion based on the adjustment ratio input from the adjustment ratio input device 3.
In addition, the stereoscopic image depth control device 4 receives a three-dimensional model from a three-dimensional model generating device (not shown), and there are no particular limitations on the type or generating method of this three-dimensional model.

[Configuration of stereoscopic image depth control device]
The configuration of the stereoscopic image depth control device 4 will be described with reference to FIG.
As shown in FIG. 3, the stereoscopic image depth control device 4 includes a depth compression section 40 and an element image generation section 44 .

The depth compression unit 40 receives a three-dimensional model of a subject and compresses the depth of the input three-dimensional model. As shown in FIG. 3, the depth compression unit 40 includes a static vertex coordinate calculation unit 41, a dynamic vertex coordinate calculation unit 42, and a vertex coordinate adjustment unit 43.

The static vertex coordinate calculation unit 41 calculates static vertex coordinates, which are the coordinates of each vertex constituting a three-dimensional model, by static depth compression that does not take into account movement of the viewing position. Here, the static vertex coordinate calculation unit 41 performs static depth compression similar to that of the conventional method. For example, the static vertex coordinate calculation unit 41 converts the vertex coordinate p of the three-dimensional model into the static vertex coordinate _p0 ' by using the above-mentioned formula (1). Then, the static vertex coordinate calculation unit 41 outputs the calculated static vertex coordinate _p0 ' of the three-dimensional model to the vertex coordinate adjustment unit 43.

The dynamic vertex coordinate calculation unit 42 calculates dynamic vertex coordinates, which are the coordinates of each vertex constituting the three-dimensional model, according to the viewing position by dynamic depth compression that takes into account the movement of the viewing position. Here, the dynamic vertex coordinate calculation unit 42 performs dynamic depth compression similar to that of the conventional method according to the viewing position input from the adjustment ratio input device 3. For example, the dynamic vertex coordinate calculation unit 42 converts the vertex coordinate p of the three-dimensional model into dynamic vertex coordinate _p1 ' using the above-mentioned formula (3). Then, the dynamic vertex coordinate calculation unit 42 outputs the calculated dynamic vertex coordinate _p1 ' of the three-dimensional model to the vertex coordinate adjustment unit 43.

The vertex coordinate adjustment unit 43 adjusts the vertex coordinates of the three-dimensional model between the static vertex coordinate p0' input from the static vertex coordinate calculation unit 41 and the dynamic vertex coordinate _p1 ' input from the dynamic vertex coordinate calculation unit 42, based on the adjustment ratio s input from the adjustment ratio input device _3. Then, the vertex coordinate adjustment unit 43 outputs the three-dimensional model after the vertex coordinates have been adjusted to the element image generation unit 44.

<Vertex coordinate adjustment>
The adjustment of the vertex coordinates by the vertex coordinate adjustment unit 43 will be described in detail with reference to FIG. 4 and FIG.
Specifically, the vertex coordinate adjustment unit 43 uses the following formula (5) to calculate the vertex coordinate p _mod1 ' of the three-dimensional model so that this vertex coordinate p mod1 ' is located between the static vertex coordinate p ₀ ' and the dynamic vertex coordinate p ₁ '. At this time, the vertex coordinate adjustment unit 43 shifts the vertex coordinate p _{mod1 '} in the direction opposite to the movement direction of the viewing position.

As shown in Fig. 4, the vertex coordinate p _mod1 ' shifts between the static vertex coordinate p ₀ ' and the dynamic vertex coordinate p ₁ ' in the x-axis and y-axis directions according to the adjustment ratio s. Specifically, the closer the adjustment ratio s is to 0, the closer the vertex coordinate p _mod1 ' is to the static vertex coordinate p ₀ ', while the closer the adjustment ratio s is to 1, the closer the vertex coordinate p _mod1 ' is to the dynamic vertex coordinate p ₁ '. Note that in Fig. 4, the projection point (i _mod1 , j _mod1 ) is the point where the vertex coordinate p _mod1 ' is projected onto the display.

As shown in Fig. 5, as the adjustment ratio s approaches 0, the distortion of shape increases while the distortion of space decreases. In the example of Fig. 5, as the adjustment ratio s approaches 0, the straight line on the right side of subject A becomes distorted into a curved shape. When the adjustment ratio s = 0, this is equivalent to static depth compression, and only distortion of shape occurs, and no distortion of space occurs.
On the other hand, as the adjustment ratio s approaches 1, the spatial distortion increases while the geometric distortion decreases. In the example of Fig. 5, as the adjustment ratio s approaches 1, the subject B slides to the right. When the adjustment ratio s = 1, this is equivalent to dynamic depth compression, and only spatial distortion occurs, with no geometric distortion.

Here, it is considered that the optimal value of adjustment ratio s differs depending on the viewing environment of the stereoscopic image display device 1. For example, in a viewing environment with a large screen such as a movie theater, the viewer 9 rarely moves his/her head while viewing. For this reason, it is better to alleviate distortion of shape rather than distortion of space, so the value of adjustment ratio s is made large. On the other hand, in a viewing environment with a small screen such as a tablet, the screen can be easily moved with both hands, so the relative positional relationship between the viewer 9's head and the display is likely to change significantly. For this reason, it is better to alleviate distortion of space rather than distortion of shape, so the value of adjustment ratio s is made small.

Furthermore, it is believed that the value of the adjustment ratio s will differ depending not only on the viewing environment but also on the characteristics of the content. For example, in content with large camera motion, it is believed that the viewer 9 will not actively move his/her head because it contains sufficient motion parallax. In this case, it is better to prioritize reducing distortion of shape, so the value of the adjustment ratio s is made large. On the other hand, in content that is viewed from multiple directions, it is believed that the viewer 9 will actively move his/her head. In this case, it is better to prioritize reducing distortion of space, so the value of the adjustment ratio s is made small. For example, by conducting visual psychology experiments using an adjustment method, etc., it is possible to determine the optimal value of the adjustment ratio s according to the viewing environment and type of content.

Returning to FIG. 3, the description of the stereoscopic image depth control device 4 will continue.
The element image generating unit 44 generates element images from the three-dimensional model input from the vertex coordinate adjusting unit 43. Here, the element image generating unit 44 can generate element images from the three-dimensional model by using a conventional method such as ray tracing.

The ray tracing method involves extending straight lines connecting each element lens constituting the lens array of the stereoscopic image display device 1 with each pixel of the element image, and obtaining color information at the intersection of the extended line with the surface of the three-dimensional model as the pixel value of that pixel. For example, the ray tracing method is described in detail in References 1 and 2 below, so further explanation will be omitted.

Reference 1: Katayama, "Method of converting 3D models into integral stereoscopic images," Japan Broadcasting Corporation, NHK Giken R&D, No. 128, July 2011. Reference 2: "Technology for generating integral stereoscopic images from 3D models," Japan Broadcasting Corporation, NHK Giken R&D, No. 123, September 2010.

The element image generating unit 44 may generate element images by a method other than the ray tracing method. For example, the element image generating unit 44 may generate multi-viewpoint images by capturing images of a three-dimensional model using virtual cameras arranged in an array, and convert the multi-viewpoint images into element images (for example, Reference 3).
Reference 3: Ikeya, "Method for generating integral 3D images using multiple viewpoint cameras," Japan Broadcasting Corporation, NHK Research and Development, No. 146, August 2014

Then, the element image generator 44 outputs the generated element image to the stereoscopic image display device 1. The stereoscopic image display device 1 then displays the element image via the lens array, allowing the viewer 9 to view the stereoscopic image T.

[Operation of stereoscopic image depth control device]
The operation of the stereoscopic image depth control device 4 will now be described with reference to FIG.
6, in step S1, the static vertex coordinate calculation unit 41 calculates the static vertex coordinates of each 3D model by static depth compression. For example, the static vertex coordinate calculation unit 41 converts the vertex coordinate p of the 3D model into the static vertex coordinate _p0 ' by using the above-mentioned formula (1).

In step S2, the dynamic vertex coordinate calculation unit 42 calculates the dynamic vertex coordinates of the three-dimensional model according to the viewing position by dynamic depth compression. For example, the dynamic vertex coordinate calculation unit 42 converts the vertex coordinate p of the three-dimensional model into the dynamic vertex coordinate _p1 ' by using the above-mentioned formula (3).

In step S3, the vertex coordinate adjustment unit 43 adjusts the coordinates of each vertex of the three-dimensional model between the static vertex coordinate _p0 ' calculated in step S1 and the dynamic vertex coordinate _p1 ' calculated in step S2, based on the adjustment ratio s input from the adjustment ratio input device 3. Specifically, the vertex coordinate adjustment unit 43 calculates the vertex coordinate p _mod1 ' of the three-dimensional model using the above-mentioned equation (5).

In step S4, the element image generation unit 44 generates an element image from the three-dimensional model whose vertex coordinates have been adjusted in step S3. For example, the element image generation unit 44 generates an element image from the three-dimensional model using a ray tracing method.

[Action and Effects]
The stereoscopic image depth control device 4 according to the first embodiment can adjust the ratio of both spatial distortion and geometric distortion so as to alleviate both of them, based on the adjustment ratio s set by the viewer 9. That is, the stereoscopic image depth control device 4 compresses the depth of the scene in order to avoid blurring that occurs when a scene with a wide depth is displayed on the stereoscopic image display device 1, and the viewer 9 can arbitrarily adjust the geometric distortion and spatial distortion that occur at that time.

Second Embodiment
With reference to FIG. 7, a stereoscopic image display system 100B according to the second embodiment will be described with respect to differences from the first embodiment.
7, the stereoscopic image display system 100B includes a stereoscopic image display device 1, a viewing position measurement device 2, an adjustment ratio input device 3, and a stereoscopic image depth control device 4B. Note that each means of the stereoscopic image depth control device 4B is similar to that of the first embodiment, and therefore a description thereof will be omitted.

[Configuration of stereoscopic image depth control device]
In the second embodiment, the stereoscopic image depth control device 4B differs from the first embodiment in that it learns the optimal adjustment ratio s. For this reason, the stereoscopic image depth control device 4B includes a depth compression unit 40B, an element image generation unit 44, a history DB storage unit 45, a learning model generation unit 46, an adjustment ratio acquisition unit 47, and a switch 48. In addition, the depth compression unit 40B includes a static vertex coordinate calculation unit 41, a dynamic vertex coordinate calculation unit 42, and a vertex coordinate adjustment unit 43B.
It should be noted that the static vertex coordinate calculation unit 41, the dynamic vertex coordinate calculation unit 42, and the element image generation unit 44 are similar to those in the first embodiment, and therefore their explanation will be omitted.

The history DB storage unit 45 stores a history database. This history database is a database that indicates the relationship between the adjustment ratio s and the depth position and viewing distance of the stereoscopic image T. For example, the history database may be a relational database.

The learning model generation unit 46 refers to the history database and generates a learning model that expresses the adjustment ratio s as a relationship between the depth position of the stereoscopic image T and the viewing distance. Here, the learning model generation unit 46 stores the learning model in a memory (not shown).

<Creating a learning model>
The generation of the learning model by the learning model generating unit 46 will be described in detail below.
For example, the learning model is a linear model that expresses the adjustment ratio s as a relationship between the depth position _t1 and the viewing distance _t2 of the stereoscopic image T, as shown in the following formula (6). Note that in formula (6), α, _β1 , and _β2 represent predetermined coefficients.

In this case, the learning model generation unit 46 only needs to find the coefficients α, _β1 , and _β2 in equation (6) as an optimization problem. Specifically, the learning model generation unit 46 solves the following equation (7) using the least squares method to find the coefficients α, _β1 , and _β2 . In equation (7), the subscript n of s _(n) , t1 _(n) , and _t2(n) represents the nth record stored in the history database (where n=0, 1, 2, ...).

Returning to FIG. 7, the description of the stereoscopic image depth control device 4B will continue.
The adjustment ratio acquisition unit 47 inputs a depth position and a viewing distance, and acquires an adjustment ratio s corresponding to the input depth position and viewing distance from the learning model. Here, the adjustment ratio acquisition unit 47 inputs a viewing distance from the viewing position measurement device 2, and the viewer 9 inputs a depth position designated in advance by the video producer. In this way, the adjustment ratio acquisition unit 47 inputs a depth position and a viewing distance corresponding to the viewing environment of the viewer 9 into the learning model, and can acquire an adjustment ratio s suitable for the viewing environment of the viewer 9 from the learning model.

The switch 48 selects either the adjustment ratio s input from the adjustment ratio input device 3 or the adjustment ratio s acquired by the adjustment ratio acquisition unit 47, and outputs the selected adjustment ratio s to the vertex coordinate adjustment unit 43B. Here, the viewer 9 can manually set which adjustment ratio s the switch 48 will select. Note that in the example of FIG. 7, the switch 48 selects the adjustment ratio s acquired by the adjustment ratio acquisition unit 47.

The vertex coordinate adjustment unit 43B adjusts the vertex coordinates of the three-dimensional model based on the adjustment ratio s input from the switch 48. That is, the vertex coordinate adjustment unit 43B adjusts the vertex coordinates of the three-dimensional model based on the adjustment ratio s input from the adjustment ratio input device 3 or the adjustment ratio s acquired by the adjustment ratio acquisition unit 47. Note that the adjustment of the vertex coordinates itself is the same as in the first embodiment, so further explanation will be omitted.

[Action and Effects]
The stereoscopic image depth control device 4B according to the second embodiment can obtain an adjustment ratio s suited to the viewing environment of the viewer 9, and can therefore automatically adjust the ratio between spatial distortion and geometric distortion in accordance with the viewing environment of the viewer 9. This allows the stereoscopic image depth control device 4B to improve convenience.

(Modification)
Although each embodiment of the present invention has been described in detail above, the present invention is not limited to this, and includes design modifications and the like within the scope of the present invention.
In the above embodiment, the learning model is described as a linear model, but is not limited to this. For example, the learning model may be learned by machine learning or deep learning.

In the above embodiment, the stereoscopic image depth control device is described as independent hardware, but this is not limited to the above. It can also be realized by a program that causes the hardware resources of a computer, such as a CPU, memory, and hard disk, to operate as the stereoscopic image depth control device. These programs may be distributed via a communication line, or written to a recording medium such as a CD-ROM or flash memory and distributed.

Reference Signs List 1: Stereoscopic image display device 2: Viewing position measurement device 3: Adjustment ratio input device 4, 4B Stereoscopic image depth control device 40: Depth compression section 41: Static vertex coordinate calculation section 42: Dynamic vertex coordinate calculation section 43, 43B: Vertex coordinate adjustment section 44: Element image generation section 100: Stereoscopic image display system

Claims (5)

A stereoscopic image depth control device that compresses a three-dimensional model of a subject in depth so that the three-dimensional model falls within a depth reproduction range that satisfies a preset resolution, and generates an element image from the three-dimensional model after the depth compression,
a static vertex coordinate calculation unit that calculates static vertex coordinates, which are the coordinates of each vertex constituting the three-dimensional model, by static depth compression without taking into account movement of the viewing position of the stereoscopic image;
a dynamic vertex coordinate calculation unit that calculates dynamic vertex coordinates, which are coordinates of each vertex constituting the three-dimensional model according to the viewing position, by dynamic depth compression taking into account the movement of the viewing position;
a vertex coordinate adjustment unit that adjusts vertex coordinates of the three-dimensional model between the static vertex coordinates and the dynamic vertex coordinates based on an adjustment ratio between the static depth compression and the dynamic depth compression;
an element image generating unit that generates the element image from the three-dimensional model after the vertex coordinates have been adjusted;
Equipped with
The static depth compression that does not take into account the movement of the viewing position is expressed by the formula (1) that converts the vertex coordinates p( x, y, z) of the three-dimensional model into vertex coordinates p ₀ ′(x ₀ ′, y ₀ ′, z ₀ ′),

The dynamic depth compression taking into consideration the movement of the viewing position is expressed by equation (3) which converts the vertex coordinates p(x, y, z) of the three-dimensional model into vertex coordinates _p1 '( _x1 ', _y1 ', _z1 '),

A stereoscopic image depth control device comprising: As shown in the following formula (5), the vertex coordinate adjustment unit:

2. The stereoscopic image depth control device according to claim 1, wherein a vertex coordinate p _mod1 ' of the three-dimensional model is adjusted between the static vertex coordinate p ₀ ' and the dynamic vertex coordinate p ₁ ' based on the adjustment ratio s. a learning model generating unit that generates a learning model that expresses the adjustment ratio as a relationship between a depth position of the stereoscopic image and a viewing distance;
an adjustment ratio acquisition unit that inputs the depth position and the viewing distance and acquires the adjustment ratio corresponding to the input depth position and the viewing distance from the learning model;
3. The stereoscopic image depth control device according to claim 1, wherein the vertex coordinate adjustment section adjusts the vertex coordinates of the three-dimensional model based on the adjustment ratio acquired by the adjustment ratio acquisition section. 3. The stereoscopic image depth control device according to claim 1, wherein the vertex coordinate adjustment section adjusts the vertex coordinates of the three-dimensional model based on the adjustment ratio inputted from an adjustment ratio input device . A program for causing a computer to function as a stereoscopic image depth control device according to any one of claims 1 to 4.

Images