JPH04108288A - Stereoscopic video device - 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 the structure of a stereoscopic camera that captures stereoscopic images in a stereoscopic television system that provides stereoscopic images.

(Conventional technology) There are many factors that cause humans to perceive 3D objects, but the main factors include binocular parallax, which is the misalignment of images caused by the difference in the position of the left and right eyes, and gazing at an object. There is a movement in which the eyes move in or out, that is, a binocular width difference.Using two television cameras, these factors are generated to capture stereoscopic images, and these images are compared to the human eyes. A 3D television system has been developed that provides a 3D sensation by presenting images independently.As a method of presenting 3D images to humans, as shown in Figure 2, two monitors are placed between the left and right eyes. There are two methods: placing each monitor independently, as shown in Figure 3, combining the images from two monitors using a half mirror, and using polarized light to separate the left and right images into the left and right eyes. As shown in Figures 4 and 5, there is a method in which left and right images are superimposed on one monitor, and the images are separated using an electronic shutter or polarized light.An example of a method using one monitor is shown in Figure 6A. Shown below.

Moreover, the waveform diagram of each part is shown in FIG. 6B. In FIG. 6A, the left camera head 1 and the right camera head 2 take images with binocular parallax. The A/D converters 3 and 4 digitize the video signals 5 and 6 output from the left and right camera heads 1 and 2 and record them in the frame memory 7.8, and the double-speed stereo converter 9 reads out these recorded signals alternately. is incorporated into the double-speed stereoscopic video signal 10, and this is sent to the D/A converter 11.
Convert it to an analog signal and send it to double-speed scan monitor 1.
Display on 1. The liquid crystal glasses 12 open and close the liquid crystal shutters in synchronization with each frame alternately displayed on the double speed scan monitor 12 using shutter signals 13 and 14 synchronized with the frames generated by the double speed stereo converter 9.

Therefore, only the left frame L is visible to the human left eye, and only the right frame R is visible to the right eye. Therefore, humans can perceive stereoscopic images due to the effect of binocular parallax.

Place a polarized shutter on the monitor screen instead of a liquid crystal shutter, or place a polarized shutter in front of the projection lens of a video projector, and switch this polarized shutter in synchronization with the left and right frames, so that the left and right lenses have polarized lenses with different polarization directions. There is also a method in which the viewer wears polarized glasses that separate the left and right images and obtain a three-dimensional image.

Regarding conventional stereoscopic viewing technology, see "Nikkei Electronics J NcL444 (1988, 4, 4), pages 205 to 223, and Journal of the Japan Society for Precision Engineering, 5/4/2/1.
988, pages 1 to 31, are discussed in detail.

[Problem to be solved by the invention]

By the way, there are the following two methods for the relative positional relationship between the two camera heads 1.2 that capture images.

The L type is such that the optical axes of the two camera heads are arranged in parallel, as shown in FIG. 2 placed in parallel
Assuming that there is an object 15 to be imaged in the hydraulic power of the camera heads 1 and 2 of the stand, an image of this object is focused on an imaging surface 18.19 by a lens 16.17. In Figure 7, the image Q of the imaging plane
, r are shown by leader lines.

The two photographed images Q and r are displayed alternately on the monitor screen 20 as shown in FIG.
This can be seen in 2. Since the left eye 21 sees the image Q seen from the left side, and the right eye 22 sees the image r seen from the right side, the position of the composite image 23 seen due to binocular parallax is in front of the monitor screen.

Now, in FIG. 7, if the object 15 moves to an infinity point on the intermediate axis of the two camera heads 1.2, the light from the object becomes parallel rays, and its images are captured by the left and right images. Side 1
8.19 are connected to each center point. When this is displayed on the monitor screen 20 shown in FIG. 8, the point at infinity will be located at the center of the monitor screen 20. In other words, the light from the object at infinity becomes a parallel ray, so there is no parallax between the aircraft and r.If the object moves left and right at infinity, the position of the 3D image will be the same as that on the monitor screen. 2
Move above 0. In other words, the position of the monitor screen corresponds to the point at infinity. Therefore, a stereoscopic image corresponding to all points in the real space will be generated in front of the monitor screen 20. Furthermore, in FIG. 7, the area inside the parallel optical axes 24.25 of the two camera heads 1 and 2 corresponds to the area inside the optical axes 26.27 of both eyes in FIG. .

Therefore, the actual space is compressed and expressed toward the front side of the monitor screen 2o, and the left and right spread is also compressed according to the distance to the object. Therefore, the image is deformed, making it impossible to accurately represent a three-dimensional image. This is shown using the formula as follows.

Define the symbols as shown in Figure 9. Point p(x,
In y), the position mL of the image on the imaging surface 18 of the left camera and the position mR of the image on the imaging surface 19 of the right camera are expressed as follows based on the distance d from the imaging surface to the lens and the distance between the lenses. be done.

y (1), mt from formula (2), = (b/2+x)
・=(3)rnR=Cx-b/2)
.=(4) However, in the settings shown in FIG. 9, mR becomes a negative value, so it is shown with a negative sign in FIG.

The sizes ML and MR of the images on the monitor screen 20 are expressed by the size S of the imaging surface and the size S of the monitor screen as follows.

ML = m
...(5) M R: -m R...(6
) In the settings shown in FIG. 9, MR also becomes a negative value, so it is shown with a negative sign in FIG.

From the above, the position of the stereoscopic image p(x,y) can be obtained as follows. Equation (10) and (13) with the distance between the eyes as E
, From formula (+4), from formula (8), (7), from formula (9), from formula (9) and (14), Esy Y= bdS+Esy (1), from formula (5), (2), (6) From equation (II), from equation (12) and from equation (15), it can be seen that it is a function of X and y and is influenced by y. Also, from equation (16), Y is unrelated to X, but it is not linearly related to y. It can be seen that when y becomes infinite, Y converges to D, and the point at infinity is reproduced at the position of the monitor screen. At this time, if X is a finite value, X becomes O regardless of X, and the points at infinity in the space on the subject side are concentrated at one point at the center of the monitor screen. When X expands infinitely, X converges to -. In this way, when the TV cameras are arranged parallel to each other, the space on the subject side is compressed to the front side of the monitor screen 20 in the stereoscopic playback side space, and the further away from the camera, the more compressed the horizontal direction becomes. be done.

Therefore, it is not possible to reproduce a correct stereoscopic image.

Figure 101. As mentioned above, a method in which the optical axes of two television cameras intersect is also used. In case 2, as will be described later, a stereoscopic image can be reproduced on the rear side of the monitor screen, but the imaging surfaces 18 and 19 are oblique.

Since they are not facing the same η direction, the following inconvenience occurs. Hold a square surface viewed from the front as the subject"-5
) If an object is used, the left camera will view the square from the left side, so as shown in FIG. shows an image with the right side enlarged and the square side reduced.
When these images are displayed superimposed on the monitor screen 201, the lengths of the corresponding vertical lines in the two images are different. If you try to view this in 3D, as shown in Figure 12, the vertical lines 28 of the 3D image to be created will be unnatural because images of different lengths will be combined, and the correct 3D effect will not be obtained. do not have.

This will be illustrated below through analysis.

As shown in Fig. 13, the distance from the line connecting the centers of the two lenses 16 and 17 to the intersection 29 of the optical axis of the camera is a,
The distance between the centers of the imaging surfaces 18 and 19 is b, and the lenses 16 and 17
If the distance between points 16 and 29 is a + e /4, and the point p
The distance between and point 3o is (e/2)x, the distance between points 29 and 30 is (e/2)x ay-, and the distance between points 30 and 32 is /v'a 2+ e 2/4 becomes. Therefore, the position m of the image on the imaging surface 18 is expressed as follows.

is expressed in

Find mR in the same way. In FIG. 14, the interval between points 29 and 33 is ay, and the interval between points 29 and 34 is (a-y).
y) V1 〒] 1-71-/a, the distance between points 33 and 34 is (a-y) (e/2)/a, the distance between points p and 34 is x-(a-y) (e/2)/a, point p Therefore, the position ImR of the image on the imaging surface 19 is as follows, since the conditions for projection onto the monitor screen are the same, the formula ~(lO) holds true.

Therefore, the position of the stereoscopic image P(X.

Y) can be found as follows.

(5).

(6).

(17).

From the formula...(19)...(20) Therefore, from the formula (10), it is 1! As a result of the above, X and Y become non-linear with respect to x and y, and distortion occurs in the reproduced image. Therefore, a correct stereoscopic image cannot be imaged.

As a result of the above, it has been found that correct stereoscopic images cannot be imaged using conventional methods. Therefore, there is a need for a stereoscopic camera that can reproduce accurate stereoscopic images using a camera with a structure different from conventional cameras.

Furthermore, when the observer moves back and forth or left and right, the 3D image changes accordingly, resulting in a distorted image, making it impossible to obtain a correct 3D image. A method is needed that will not change the image even when

[Means to solve the problem]

In order to achieve the above object, in the stereoscopic camera according to the present invention, the two television cameras are arranged so that the normal lines of the imaging planes are parallel to each other, and the distance between the two imaging planes is smaller than the distance between the centers of the two lenses. Signal processing is performed to increase the distance between the imaging points corresponding to the center of the image, so that the line connecting the center of the lens and the imaging point corresponding to the center of the image intersects in front of the three-dimensional camera.

[Effect]

As shown in Fig. 15, two television cameras are arranged in parallel, and the straight line connecting the point at a distance a in front of the two television cameras and the center of the lens of the television camera is the imaging plane 18.1.
The point 36.37 that intersects 9 is set as the center point of the image, and the image is reproduced so that the center point 36.37 of the image coincides with the center point 38 of the monitor screen 20. At this time, point p(
The position of the three-dimensional reconstructed image P(X, Y) of x, y) is determined as follows. From the geometrical relationship shown in FIG. 15, the distance between the image center points is determined as follows.

d a Therefore, the image position mLt mR is (e/2+x)-(e/2) mL ad
y...(25) d Y Also, from the size S of the imaging surface and the size S of the monitor screen,
The size of the images ML and MR on the monitor screen 20 are M = ... (27) M R = -m R ... (28) The position of the stereoscopic image p (x, y) is as follows. It can be obtained by

D (ML+MR)/2 Y x (29) Y-DMR-ML (30) From formula (30), 2 d E S x / m y (29), from formula (31) (25) , From formula (26), mL+mR: (2(e/2+x)-e) (27),
From equations (28), (33), and (34), 22 shows that the ratio of the distance a from the TV camera lens to the intersection point 29 of the optical axis and the distance d from the lens to the imaging surface is the camera distance e and the observer is equal to the ratio of the binocular distance E multiplied by the ratio of the monitor screen size S to the imaging surface size S. That is, eS -=-・-...(38) Es Substituting equation (38) into equation (37), X = -x
...(39), and from equations (31) and (36), (32),
Substituting equation (38) into equation (40) from equations (35) and (36), S e Here, the distance from the observer to the monitor screen and the distance d from the lens to the imaging surface are determined by the size of the monitor screen. Suppose that it is equal to the ratio of the size S and the size S of the imaging surface.

That is, Substituting equation (42) into equation (41) yields.

Y=-y...
(43) From equations (39) and (43), the three-dimensional reconstructed image p(
x, y) is expressed E/e times the original point p and is completely isotropic. Therefore, an accurate stereoscopic image without distortion can be obtained. If the camera interval e is made equal to the distance between the observer's eyes, the stereoscopic reconstructed image can accurately represent the environment in front of the stereoscopic camera as it is. Furthermore, if a three-dimensional camera with a small camera interval e is used while maintaining the conditions expressed by equations (38) and (42), the reconstructed image will be enlarged.

Conversely, if a camera with a large e is used, the three-dimensional reconstructed image will be reduced. At this time, linearity is maintained.

Let E/e be the stereoscopic image magnification rate ν.

y=-river (44) From equations (38), (42), and (44), a=D/ν
...(45) Therefore, the two conditions for obtaining a distortion-free stereoscopic image are expressed as follows from equations (42) and (45).

[Condition 1] The observer observes from a distance from the monitor screen equal to the distance from the center of the camera lens to the imaging surface multiplied by the ratio of the monitor screen size S to the imaging surface size S. thing.

[Condition 2] The distance a from the intersection of the optical axes of the two cameras to the line connecting the optical centers of the camera lenses is the distance from the observer to the monitor screen, the camera distance e, and the distance between the observer's eyes E. The ratio with
In other words, it must be set to be equal to the 3D magnification multiplied by the reciprocal of the 3D magnification γ.

〔Example〕

The first method of shifting the center of the image on the imaging surface is shown in FIG.
On the imaging surface 39, if the interlace method is used, the imaging surface is scanned every other scanning line, and a video signal 40 is obtained. In this video signal 40, a horizontal or vertical synchronizing signal 41 is replaced with the video signal of this portion between the next scan and the next scan. On the monitor screen 42, the screen is scanned after a certain time t has elapsed since the synchronization signal, and the video signal is changed to brightness and darkness.At this time, the center 43 of the imaging surface almost coincides with the center 44 of the monitor screen. There is. now.

If a synchronization signal 45 that is earlier than usual is combined with the video signal 40, the video signal that is earlier than usual will be scanned onto the monitor screen 42, so the video in the screen frame 46 will be displayed on the monitor screen 42. It turns out. Therefore, the image moves to the right on this monitor screen 46. As a result, the image center 47 of the screen frame 46 corresponds to the image center 48 located to the left on the imaging surface 39. If the synchronization signal is delayed, one image will move to the left, and the center of the image on the imaging surface 39 will shift to the right. It is possible to adjust the position of the image center.

The stereoscopic composite signal 48 is as shown in FIG. 16, in which signals from the left and right cameras are alternately combined. The range indicated by L is the video signal of the left camera, and the range indicated by R is the video signal of the right camera. The horizontal synchronization signal is shifted in the opposite direction between the left camera and the right camera with a wide vertical synchronization signal as the boundary, and the horizontal scanning signal 49 on the monitor is a horizontal scan signal after a certain period of time tl based on the horizontal synchronization signal. start. Further, the first horizontal scan after the vertical synchronization signal starts the horizontal scan after a time tz from the vertical synchronization signal.

At this time, the relationship between tl and tz is determined so that the period of the sawtooth wave for horizontal scanning is constant. Note that the color burst signal of the horizontal synchronization signal is omitted in the figure.

FIG. 17 shows an example of shifting the synchronization signal of a stereoscopic camera, and the synchronization signal generator 50 is shown in FIG. 18 (Ca).

Three kinds of phase-shifted /</</response signals shown in (b) and (c) are generated. Synchronization signal CL output from the left camera 51 and right camera 52 synchronization signal generators 50.

Video signals 53 and 54 are generated in synchronization with CR and input to a synchronization signal synthesizer 55 and 56. Synchronous signal synthesizer 55.5
6 synthesizes the video signals 53 and 54 and the synchronization signals SL and SR output from the synchronization signal generator 50, respectively, and outputs one composite video signal 57.58. The stereoscopic synthesizer 59 alternately combines the composite video signals 57 and 58 to create a stereoscopic video composite signal 6o, and outputs this. Here, the synchronization signal generator 5o generates synchronization signals CR and CL of the television camera (b
) to move the center of the image to the left side 1
If you want to send the pulse signal (a) with a phase lead than the pulse signal (b) as the synchronization signal SL or SR, and want to move the center of the image to the right, send the pulse signal (a) with a phase lag than the pulse signal (b). The pulse signal (c)
Just send it as R%,%. By adjusting the phase difference jp, t*, the position of the center of the image can be arbitrarily adjusted.

A second method for shifting the center of the image is to change the time t from the horizontal synchronization signal to the start of scanning on a monitor that reproduces stereoscopic images. As shown in FIG. 19, the synchronization signal is incorporated in the stereoscopic composite signal 61 at the same timing as usual, but the time t from the horizontal synchronization signal to the start of scanning with the horizontal scanning signal is a different time jL.
+ jR. That is, in synchronization with the left and right video signals, time 11. and tR are switched alternately. Therefore, as a result, the horizontal scanning signal 62 can be made the same as the horizontal scanning signal 49 obtained by the first method.

FIG. 20 shows an embodiment of this second method, in which a synchronizing signal generator 50 generates synchronizing signals C
Send R, CL. Although not shown, a television camera 51.5
The second video signal is combined with a synchronization signal in the usual manner, and further sent to a monitor as a stereoscopic composite video signal. The synchronization signal generator 50 is also a horizontal scan signal generator 6 of the monitor.
3 to generate a sawtooth wave 64, which is amplified by a horizontal output device 65 and drives a horizontal coil 66. The synchronizing signal generator 50 is shown in FIGS. 21(a) and 21(b).
, generate phase-shifted pulse waves shown in (C), and alternately output (b) as synchronization signals CR and CL to the television camera 51 and 52, and output S to the horizontal scan signal generator 63,
outputs (a) or (c) alternately in synchronization with CR and Ct.

Therefore, the video signals of the left and right TV cameras correspond to the horizontal synchronization signal (a) or (c), and the video centers of the left and right TV cameras have a phase shift of 1-. 1- allows the position of the image center to be changed individually.

FIG. 22 shows another embodiment, and FIG. 23 shows waveforms at various parts. This method is a method of adjusting the timing of a horizontal scan signal based on a stereoscopic composite video signal 67 sent from a television camera. A synchronization signal separator 68 extracts a horizontal synchronization signal 69 and a vertical synchronization signal 70 from the stereoscopic composite video signal 67.
Separate.

The horizontal synchronization signal 69 and vertical synchronization signal 7o are input to signal delayers 71 and 72, and are respectively delayed and input as horizontal synchronization signals 73 and 74 to the horizontal scan signal generator 63 to generate a sawtooth wave 64. The signal is amplified by the horizontal output device 65 to drive the horizontal coil 66. Delay time from horizontal synchronization signal1. and to are switched each time the vertical synchronization signal 70 is input to the signal delay device 71, so the delays of the left and right video signals and the horizontal synchronization signal correspond.

The delays tsv and ts from the vertical synchronizing signal may be determined so that the pulse interval t is constant.

A third method for shifting the center of the image is to use an image memory. A/D converts video signals obtained from two TV cameras
When the video signal is converted into digital data and input to the image input circuits 75 and 76 as shown in FIG.
Recorded at 7.78. The stereoscopic image synthesis/reading circuit 7g alternately reads the video signals of the two image memories 77 and 78 to synthesize a stereoscopic video signal, but at this time, the images are shifted by changing the reading start address of the image memory. For example, when advancing a video signal, the beginning part of the image memory is omitted and reading starts from the middle. When delaying the video signal, the video signal can be delayed by starting reading before the end of the image memory and returning to the start address when the end point is reached, or by adding a dummy signal to the start.

Now, let's consider changing the magnification ratio of a 3D image.

From equations (39) and (43), the magnification rate of the reproduced stereoscopic image is determined by the ratio of the distance E between the observer's eyes and the distance e between the two television cameras. Since the distance E between the observer's eyes is constant depending on the observer, the distance e between the two television cameras can be changed in order to change the magnification ratio.

At this time, the two conditions mentioned above must be observed.

The condition expressed by equation (42) has no direct relation to the change of e and is maintained as is. The condition expressed by equation (45) indicates that if E and D are constant, the ratio of a and e must be constant.

According to equation (24), unless d is changed, the amount of deviation from the center of the image (b
-e)/2 If e is changed while keeping constant, a will also change proportionally. Therefore, if you want to enlarge a stereoscopic image, you can reduce the distance e between the two television cameras, and if you want to reduce it, you can increase e. An embodiment of the means for adjusting the distance e between the cameras is shown in FIG. 25. Two television cameras 80 and 81 are fixed on slide bases 83 and 84 supported by a slide rail 82.

This slide base is slid in the opposite direction by means of a screw 4I85 having opposite threads at both ends and a female screw 86°87 which engages with this screw 4I85. The threaded rod 85 is driven by a motor 88.

By the way, the position of the viewer relative to the monitor screen is closely related to the distance between the TV cameras, the size of the screen and imaging surface, and the focal length of the lens, so in order to see the correct 3D image, you must observe it from a designated location. However, being constrained to a certain position means losing freedom of movement, which is often undesirable. Therefore, we will consider what can be done to ensure that the 3D image remains unchanged when the observer moves from its normal position.

Now, as shown in Figure 26, change the observation point from the monitor screen to D.
Suppose that the object moves from a distance of D' to a distance of D'. The position of the image on the monitor screen [ML? Assuming that MR remains the same, the stereoscopic image P moves backward in the direction away from the monitor screen, as shown by the intersection of the broken lines. Although not shown, if P is in front of the monitor screen, Move it toward you, away from the screen. That is,
The front-rear axis is extended around the monitor screen, resulting in a different magnification ratio from the left-right axis. To spin this, it is necessary to change the lengths a and d to di, as shown in Figure 27. At this time, the relationships shown in equations (42) and (45) must be maintained, that is, the values of aJ and di are (
44), (45), and (42) as follows.

/ == DI
...(46) d'=-D'
...From equations (47) and (38), e# E
If I 898 is constant.

a' is proportional to d'.

Next, consider the case where the observer moves sideways. Now, as shown in FIG. 28, if person IIJ1 moves to the right by L, the stereoscopic image P moves to the left with the screen as the fulcrum. Although not shown, if P is in front of the screen, it will move to the right. In order for this to appear in its original position, ML,
MR must move to Ml,′,MR′.

In order to prevent the stereoscopic image P from moving, as shown in FIG. 29, the lens and imaging surface may be moved laterally while keeping the position of the intersection 29 of the optical axes of the stereoscopic camera unchanged. Now suppose that both lenses move by Ω to the right. At this time, the image center points 36 and 37 are 36', 37
', the moving distance QS can be obtained as follows. From the geometrical relationship shown in Figure 29, Qs
a+cl Q a (48) Therefore, the moving distance Q of the position of the image of p(x, y) on the imaging plane 18.19 is as follows.

f1my+d y ...(50) therefore.

y+ci Q,=Q・
(51) The moving distance QH of the image on the monitor screen 20 is determined as follows. First, in FIG. 29, the center point 36.37 of the 0 image assumes a straight line 89 that divides the distance between the imaging surface 18° 19 and the monitor screen 20 into the ratio of the size S of the imaging surface and the size S of the monitor screen. Or, the intersection of the line connecting 36', 37' and the center point 38 of the monitor screen and the straight line 89 is 9
0.91 or 90'91', the straight line connecting the position of the image on the imaging plane and the corresponding position of the image on the monitor image passes through the intersection 90.91 or 90', 91'. 9
The distance #ρC between 0 and 90' or 91 and 91' is obtained as follows.

Qc 5 QsS+s ・(52) Substituting equation (49) into equation (52), S
a 0 Using the S+s formula (53) from the relationship shown in Figure, S+s Qc S+ Here, 3rd Q, +0M 12C+QM Therefore, (51).

...(54) Qs" Qm- Here, as shown in FIG.
(X', Y'), then ((ML - RM) + (M
R-QM))/ 2- LX' L Y' MR-MLD-Y Y ... (63) ... (56) From formula (63), (E/2-L) + (E/2+L) Y '...(57) From equation (57), D (MR-ML,)Y' = E (D - Y'...(58) Substituting equation (64), therefore, MRML+E (59).

From equation (64) From equation (56) Y′ =Y ...(67) Then, It can be seen that Y does not change.

Also, D (62).

From equation (63) (59) Substituting Eq. Mt, +MR= DX ...(68) (60).

Using equations (65) and (68), one of the positions in the X-axis direction and the position of the original stereoscopic image P(X, Y) are determined in step 11.

2DX-2(QM+L)Y DX D D here (43).

Substituting equation (55), we get Q = ...(73) Assuming that, X′ -X=0 ...(74) Then, is always equal to X.

in this way do, (42).

If the condition shown in equation (73) is satisfied in addition to the condition shown in equation (45), the position of the image will not change even if the observer moves sideways.

The condition for equation (73) is based on equation (45). It can also be written as follows.

Substituting equation (42), we get: The ratio of the lateral movement distance Q of the observer to the lateral movement distance Q of the camera lens can be set to be equal to the stereoscopic image magnification factor ν.

In addition, in order to ensure that the intersection point 29 of the television camera's light...(71) axes does not change even when the camera lens moves laterally, and by substituting equation (45).

Of course, the image center points 36 and 37 must be laterally moved by Qs shown in equation (49).

Summarizing the above, in order to prevent the position of the 3D image P (X, Y) from moving relative to the monitor screen even if the IR observer's position moves from the normal position, the following operation can be performed. .

(1) When the distance 111D between the observer and the monitor screen changes, adjust the lens so that the rate of change a'/a of a and the rate of change d'/d of d are equal to the rate of change D'/D of D. change the focal length d of the camera and move the camera back and forth.6 (2) When the observer moves a certain amount in the lateral direction, the lens of the television camera changes by the value a obtained by dividing L by the stereoscopic image magnification rate. move sideways. Furthermore, the position of the center of the image on the imaging plane must be moved laterally by a factor of Q with respect to the lens.

One possible method for automatically making such adjustments is as follows. First, changes D' and L in the observer's position
An example of a method for measuring is shown in #32@. An observer 92 wears a helmet 94 equipped with a magnetic sensor 93. A saucer 96 that generates an artificial magnetic field is attached to the monitor 95, and the magnetic sensor 93 receives the artificial magnetic field of the saucer 96 and outputs a signal according to its position. From this output, the position of the magnetic sensor 93 with respect to the saucer 96 can be detected, and D' and L are the digits 1 and 1 of the 3D camera 97 that can be detected by this magnetic sensor. The means for adjusting a' and a is a robot arm 9 as shown in FIG.
8 is used, but an XY child table may be used.

An embodiment of the three-dimensional camera 97 is shown in FIGS. 34 and 35. Two television cameras 98 and 99 are connected to the slide lever 1.
Slide base 102 supported by 00,101.
It is installed on top of 103. Slide base 102
, 103 are engaged with screw studs 106, 107, and the drive motors 108,
By driving the screw studs 106 and 107 by 109, the positions of the slide bases 102 and 10'3 can be changed independently. Note that the sensor 110 that detects the position
, 111 are threaded by gears 112, 113.
07, and it is possible to position the slide bases 102 and 103 at specified positions by means such as a one-position servo.

Also, on the front surface of the television camera 98, 99 is provided a spacing adjustment means similar to that shown in FIG. 22, but this spacing adjustment means only adjusts the spacing between the lenses. The lenses used here are electric telephoto lenses 114 and 115. The distance between the lens 114 and the lens 115 can be adjusted by a screw stud 85 having a reverse thread.

Magnetic sensor control device 1 whose control block is shown in FIG.
16 outputs a signal indicating the distance D between the observer and the screen and the lateral movement distance from the center line of the monitor screen,
It is input to the arithmetic and control unit 117. Also, keyboard 1
From step 18, the distance E between the eyes of the observer and the magnification factor Y of the stereoscopic image are input to the arithmetic and control unit 117. The arithmetic control block [117] uses the arithmetic block 119 to calculate e from the signals E and ν, and outputs this to the controller 120 of the distance adjustment device for the radio telegraph camera to set the camera distance to e. Arithmetic block 12
1, a is determined from the D and ν signals and output to the robot control device 122. In addition, the calculation block 123 has L and ν
Then, ρ is determined and output to the robot control device 122.

As a result, the robot 98 moves the position of the stereoscopic camera 97 to the specified value of a, 12. In addition, the calculation block 124 calculates the distance d from the optical center of the lens to the imaging surface based on the value of D, the image sensing surface size S and the monitor screen size S, which are the default settings, and calculates the distance d from the optical center of the lens to the imaging surface. 126 and adjust the optical system. The calculation block 127 calculates the moving distance QS of the imaging surface from the data of a, d, and Q, and calculates the position of the imaging surface rll X RtXL from this result and the value of e, respectively.
8 and 129, this is input to the position controllers 130 and 131 of the left and right television cameras, and the television cameras are moved to the designated positions.

As described above, even if the distance between the eyes of the observer is different, the distance between the lenses of the TV camera can be adjusted to suit the person, the magnification ratio ν of the stereoscopic image can be changed, and the distance between the eyes of the observer can be changed. It is possible to prevent the 3D image from being distorted or moved even when

〔Effect of the invention〕

According to the present invention, since the video signals of the two cameras can be shifted in parallel, accurate stereoscopic viewing without distortion can be achieved.

Further, even if the observer is at a position deviated from the normal position, a correct stereoscopic image can be obtained.

[Brief explanation of drawings]

FIG. 1 is an explanatory diagram of a video signal according to an embodiment of the present invention, FIGS. 2, 3, 4, 5, 6A, and 6B.
Figure 7 is a horizontal cross-sectional view of a conventional stereoscopic camera, Figure 8 is a horizontal cross-sectional view of a stereoscopic reproduction system, and Figure 9 is a conventional stereoscopic optical system. Explanatory drawings, FIG. 10 is another principle diagram of conventional stereoscopic vision, FIGS. 11 and 12 are explanatory diagrams of images of conventional technology, FIGS. 13 and 14 are diagrams of conventional optical system of stereoscopic vision, Fig. 15 is an optical explanatory diagram of the present invention, Figs. 16 to 18 are explanatory diagrams of video signals of one embodiment of the invention, and Figs. 19 to 21 are explanatory diagrams of video signals of other embodiments of the invention. 22 and 23 are video signal explanatory diagrams of another embodiment of the present invention, FIG. 24 is a video signal explanatory diagram of another embodiment of the present invention, and FIG. 25 is a video signal explanatory diagram of another embodiment of the present invention. 26 to 31 are optical explanatory diagrams, FIGS. 32 and 33 are partial external views of the embodiment of the third invention, and FIG. 34 is the embodiment of the third invention. A cross-sectional view of
FIG. 35 is a plan view of an embodiment of the third invention, and FIG. 36 is a control block diagram of the embodiment of the third invention. 16.17...Optical center of lens, 18.19...
Imaging plane, 2o...Monitor screen, 21.22...Observer's eye position, 4o...Synthetic video signal for stereoscopic viewing, 4
1゜45...Synchronization signal, 49...Horizontal scan signal, 50...Pulse signal generator, 71.72...Signal delay device, 77.78...Image memory, 79...Stereoscopic Image synthesis readout circuit, 83, 84, 102, 103・
・Slide base, 85, 106, 107 ・Screw attachment. 93 - Magnetic sensor, 96 - Saucer, 98 - Robot, 114, 115 - Zoom lens, 116 - Magnetic sensor controller, 117 - Arithmetic control unit. 118...Keyboard, 119, 121, 123゜12
4.127, 128, 129... calculation block, 12
2. Robot controller, 125, 126... Zoom lens controller, 130, 131... Television camera position controller, tt, tL, tn... Delay time from horizontal synchronization signal, t2... From vertical synchronization signal Delay time, tn...Delay time, t...Advance time. 1st mouth

Claims (1)

[Claims] Two types of images with parallax taken by one or two cameras are played back on the screen of a video playback means, and among the images on the screen,
In a stereoscopic video system in which an observer's left and right eyes can view stereoscopic images due to parallax by observing only the images from the respective cameras among the cameras, the optical axes of the two cameras 3. A stereoscopic imaging device, characterized in that the cameras are arranged so that the images are parallel to each other, and electronic circuit means is provided for moving the center of the reproduced image from side to side. 2. The stereoscopic video apparatus according to claim 1, wherein the electronic circuit means is a time adjustment circuit that advances or delays the timing when combining the horizontal synchronization signal with the video signal. 3. The stereoscopic video apparatus according to claim 1, wherein the electronic circuit means is a time adjustment circuit that advances or delays the time from receiving the horizontal synchronization signal to starting the horizontal scan signal. 4. A three-dimensional image device according to claim 1, wherein the electronic circuit means is a three-dimensional image reading circuit capable of changing a start address when reading a video signal from an image memory. 5. Consisting of an interval adjustment means for the two television cameras, a control means for the interval adjustment means for adjusting the interval between the two television cameras to a specified value, and an input means for the magnification rate of the stereoscopic image,
3. A stereoscopic video apparatus, characterized in that the control means uses the interval adjustment means to adjust the interval between two television cameras according to the input magnification value. 6. Consisting of arbitrary detection means for the observer's head, means for moving the imaging planes of two television cameras, two electric zoom lenses, means for moving the entire camera system, and calculation control means,
The calculation control means adjusts the position of the camera, the position of the imaging surface, and the zoom rate in accordance with the head position signal inputted from the head position detection means of the observer, so that the stereoscopic image is not distorted or moved. A stereoscopic imaging device that is characterized by eliminating the need for