RU2340116C1 - Method of playing back stereoscopic images from camcorder with converging axes - Google Patents

Abstract

FIELD: physics.

SUBSTANCE: images from camcorders are played back on first and second screens, installed in vertical planes and optically interconnected using a beam splitter. The first and second screens are arranged symmetrically relative the beam splitter. At least one of the screens is turned about a vertical axis, passing through the centre of the screen, providing for observation of images from the first and second screens in different planes, intersecting at an angle 2a. Angle a is determined from the relationship:

where V is the distance from the centre of the first screen to the observer; M is the increase in the frame, with which the image is played on the screen; f is the focal distance of objects of the camcorder; β is the angle of convergence of the optical axes of the objects of the camcorder.

EFFECT: higher quality of stereoscopic images.

3 cl, 4 dwg

Description

The invention relates to methods for reproducing stereoscopic television and video images (hereinafter referred to as stereoscopic images) and can be used in such fields as science, education, medicine, manufacturing, including microscopy, endoscopy, telemedicine, underwater television, where real-time high-quality volumetric production is required images of objects.

In stereoscopic systems, the volume of the observed image is achieved by presenting the viewer with two flat images (stereopairs) corresponding to the left and right viewing angles of a three-dimensional scene, in which each eye of the viewer sees only the image of the corresponding angle. A stereo pair is obtained using two cameras, the optical axis of the lenses of which are either parallel to each other, or converge in the center of the studied scene.

A scheme with converging axes has several advantages, however, in the known methods for reproducing stereoscopic images when using a stereo pair from cameras with converging axes, the quality of the reproduced stereoscopic image is not high enough (see Valius N.A. Stereoscopy. M .: Publishing House of the USSR Academy of Sciences. 1962. 379 pp. Woods A., Docherty T., Koch R. Image Distortions in Stereoscopic Video Systems // SPIE Proc., Stereoscopic Displays and Applications IV. 1993. V. 1915. P.36-48):

1) firstly, there is a distortion of the stereoscopic image, called a curvature of the screen plane; with this distortion, objects located in the center of the screen appear to be closer to the observer than objects at the edges of the screen;

2) secondly, the vertical parallax present in the stereo pair received from cameras with converging axes prevents the perception of the stereoscopic image outside the central part of the screen.

These shortcomings can lead to a false impression about the spatial arrangement of objects in the observed scene and to fatigue of vision, and with large values of vertical parallax, to the disappearance of the stereo effect.

A known method of reproducing stereoscopic images in which a video signal containing stereopairs of images is processed using a special device and the image fragments for the left and right eyes (hereinafter, the left and right images) are displayed on the screen so that these fragments alternate with each other on the screen . Provide orthogonal linear polarization of the light fluxes coming from the left and right images, placing a micropolarizer matrix in front of the screen, and observing the image on the screen through glasses with analyzers (see US patent No. 6593959, IPC H04N 15/00).

This method is characterized by the insufficient quality of the reproduced stereoscopic image due to a twofold decrease in resolution, as well as the curvature of the screen plane and the difficult perception of the stereoscopic image outside the central part of the screen when reproducing stereo pairs obtained by shooting with converging axes.

There is a method of reproducing stereoscopic images, in which the left and right images are alternately displayed on the display screen, a liquid crystal modulator is placed in front of the screen, the modulator is switched between the first and second polarization states in synchronization with the output of the left and right images, the image on the screen is watched through glasses with analyzers (see US patent No. 6975345, IPC H04N 13/04).

This method allows you to get a stereoscopic image without lowering the resolution, however, it places higher demands on the display, which should display frames at double frequency. In addition, when reproducing stereopairs obtained by shooting with converging axes, the resulting stereoscopic image is characterized by insufficient quality due to the curvature of the screen plane and the perception of the stereo effect outside the center of the screen.

Closest to the claimed method is a method for reproducing stereoscopic images in which images of the left and right angles of a three-dimensional scene are reproduced on two screens mounted at an angle of 90 ° to each other and optically connected to each other using a translucent mirror that is positioned at an angle of 45 ° to to each of the screens. A polarizer is placed in front of each of the screens and the left and right images are observed in the plane of one of the screens through glasses with analyzers (see US patent No. 2845618, NKI: 340-369).

This method allows you to observe the image without lowering the resolution and do not impose increased requirements on the displays associated with the need to work with doubled frame rate. However, when reproducing stereopairs obtained by shooting with converging axes, the observed stereoscopic image is characterized by insufficient quality due to the curvature of the screen plane and the perception of the stereo effect outside the center of the screen.

The objective of the invention is to develop a method for reproducing a stereoscopic image, allowing to reproduce a stereo pair obtained from cameras with converging optical axes, as a stereoscopic image with undistorted reproduction of planes, freely observed both in the center and outside the central part of the screen.

The technical result of the claimed invention is to improve the quality of stereoscopic images obtained in stereoscopic television and video systems using shooting with converging optical axes of the lenses of video cameras.

This object is achieved in that when reproducing a stereoscopic image from video cameras with converging optical axes, the image from which is reproduced on the first and second screens mounted in vertical planes and optically coupled using a beam splitter, the first and second screens being symmetrically relative to the beam splitter plates, according to the proposed solution, rotate at least one of the screens around a vertical axis passing through the center crane, providing surveillance images from the first and second screens in different planes intersecting at an angle 2α, the angle α is determined from the relation

where V is the distance from the center of the first screen to the observer;

M - magnification of the frame with which the image is displayed on the screens;

f is the focal length of the lenses of the cameras;

β is the angle of convergence of the optical axes of the lenses of the cameras.

In the first embodiment of the method, one of the screens is rotated by an angle 2α, and in the second embodiment of the method, the first and second screens are rotated in the same direction by an angle α.

As screens use, for example, screens of displays with cathode ray tubes, or liquid crystal displays, or plasma displays, or displays on light emitting diodes.

In known stereoscopic systems, images for both eyes are presented to the viewer in one common plane, which corresponds only to the shooting geometry with parallel axes of the cameras, where the sensors of the cameras are also in the same plane. When shooting with converging axes, the sensors of the cameras are in different planes and when playing images from both cameras in the same plane, the observed stereoscopic image is distorted. In the inventive method, it is proposed to reproduce images from video cameras with converging axes in different planes, which would correspond to the geometry of shooting with converging axes. At a certain value of the angle 2α between these planes, set according to the proposed method, the quality of the obtained stereoscopic image increases, namely, the curvature of the screen plane in the stereoscopic image is eliminated and the negative effect of vertical parallax on the perception of the stereo effect outside the central part of the screen is eliminated.

The invention is illustrated by drawings, where in Fig.1 shows a diagram of a stereo pair from cameras with converging axes, Fig.2 is a diagram of a stereoscopic image according to the claimed method; figure 3 is a diagram of a stereo pair of a rectangular lattice from cameras with converging axes; figure 4 - representation of the left and right images of the lattice in the common plane of observation.

The positions in the drawings indicate: 1 - object; 2, 3 - video cameras; 4 - beam splitting plate; 5, 6 - positions of the first and second screens before turning; 7, 8 - centers of the first and second screens; 9, 10 - position of the first and second screens after rotation; 11 - viewer; 12, 13 - lenses of video cameras; 14 is an image of a screen placed at position 10 observed by the viewer in the light reflected from the beam splitter plate; 15 - plane convergence of cameras; 16, 17 - sensors of video cameras; 18, 19 - left and right images obtained by video cameras; 20 is a common plane of observation of the left and right images.

To improve the quality of reproduced stereoscopic images according to the proposed method, the left and right images of object 1, obtained from cameras 2 and 3 with converging axes, are reproduced on the first and second screens mounted in vertical planes and optically connected with each other using a beam splitter plate 4. In this case the left image is reproduced, for example, on the first screen, and the right image is reproduced, for example, on the second screen. The first and second screens are placed symmetrically with respect to the beam splitter plate 4, for example, the first screen is in position 5, and the second screen is in position 6. Then, the first screen is rotated around an vertical axis passing through its center 7 by an angle α, as a result, the first screen appears in position 9. The second screen is rotated around a vertical axis passing through its center 8 by an angle α, as a result, the second screen is in position 10. In this case, both the first and second screens are rotated in the same direction, and the rotation angle α is determined elyayut from relation (1):

where V is the distance from the center 7 of the first screen to the viewer 11;

M - enlargement of the frame with which the image is displayed on the screens;

f is the focal length of the lenses 12 and 13 of the cameras 2 and 3;

β is the angle of convergence of the optical axes of the lenses 12 and 13 of the cameras 2 and 3.

The viewer 11 observes the image from the first screen at position 9 in the light transmitted through the divider 4, and the image from the second screen at position 14 in the light reflected from the divider 4, while the images are separated from the first and second screens by any known method. Thus, the viewer observes the left and right images of the object in different planes intersecting at an angle of 2α, due to which the curvature of the plane of the screen is eliminated in the observed stereoscopic image, and the stereo effect is freely perceived in the entire area of the screen.

Let us explain in more detail how the implementation of the proposed method eliminates distortion in a stereoscopic image. To do this, we first consider the causes of distortion in a stereoscopic image from cameras with converging axes when observing the left and right images of an object in the same plane. Let a rectangular lattice be located in the convergence plane of 15 cameras 2 and 3, with the optical axes of the lenses 12 and 13 of the cameras converging in the center of the lattice. Since the sensors 16 and 17 of the cameras 2 and 3 are not conjugated to the convergence plane 15 in the optical systems of the lenses 12 and 13, the left 18 and right 19 lattice images are obtained with variable linear magnification. When combining such images in the common observation plane 20, their conjugate points are characterized by an alternating horizontal parallax P _H and a nonzero vertical parallax P _V. With separate binocular observation of images 18 and 19 in a common plane 20, the presence of parallaxes P _H and P _V leads to the appearance of a variable horizontal and nonzero vertical disparity in the images obtained on the retinas of the left and right eyes of the viewer. This horizontal disparity does not correspond to the distribution of the lattice points in depth and therefore leads to a distortion of the observed stereoscopic image of the lattice. Nonzero vertical disparity makes it difficult to perceive the left and right lattice images as a single volumetric image. Such distortions occur not only for the convergence plane, but also for other planes along the entire depth of the reproduced stereoscopic image.

When observing the left and right images in different planes, provided according to the claimed method, in the images obtained on the retinas of the left and right eyes of the viewer, the geometric distortions that occur during shooting due to the discontinuity of the sensors of the cameras are compensated. As a result, the horizontal disparity of the retinal images is brought into correspondence with the distribution of the points of the object in depth, and the vertical disparity is eliminated. Thus, the implementation of the proposed method eliminates both of the reasons why in the known methods there are distortions when reproducing stereoscopic images from cameras with converging axes.

стереоскопическое изображение для наблюдения с дистанции V=60 см, имея левое и правое изображения объекта, получаемые от видеокамер с фокусным расстоянием о6ъективов f=0,35 см и углом конвергенции β=5°. Для этого левое и правое изображения объекта воспроизводят с увеличением кадра М=50 на первом и втором экранах, установленных под прямым углом друг к другу и оптически связанных между собой с помощью светоделительной пластины, расположенной под углом 45° к каждому из экранов. Затем первый и второй экраны поворачивают по часовой стрелке вокруг вертикальных осей, проходящих через их центры, причем угол поворота α определяют из соотношения (1)The inventive method can be carried out, for example, as follows. Let it be necessary to reproduce quality

a stereoscopic image for observation from a distance of V = 60 cm, having a left and right image of an object obtained from cameras with a focal length of about 6 lenses f = 0.35 cm and a convergence angle β = 5 °. To do this, the left and right images of the object are reproduced with increasing frame M = 50 on the first and second screens installed at right angles to each other and optically connected with each other using a beam splitter plate located at an angle of 45 ° to each of the screens. Then the first and second screens are rotated clockwise around the vertical axes passing through their centers, and the rotation angle α is determined from the relation (1)

The image from the first screen is observed in the light transmitted through the divider, and the image from the second screen is reflected in the light reflected from the divider, while in any known manner, the left and right images are separated. Thus, the viewer sees the left and right images in different planes intersecting at an angle of 2α, which according to the invention allows him to observe a stereoscopic image of the object without curving the planes, and the stereo effect is perceived both in the center and outside the central part of the screen.

Claims (3)

1. A method of reproducing a stereoscopic image from two cameras with converging optical axes, the image of which is reproduced on the first and second screens mounted in vertical planes and optically interconnected using a beam splitter plate, the first and second screens being arranged symmetrically relative to the beam splitter plate, characterized by turning at least one of the screens around a vertical axis passing through the center of the screen, providing an image observation d from the first and second screens in different planes intersecting at an angle of 2α, and the angle α is determined from the relation:

where V is the distance from the center of the first screen to the observer; M - enlargement of the frame with which the image is displayed on the screens; f is the focal length of the lenses of the cameras; β is the angle of convergence of the optical axes of the lenses of the cameras. 2. The method according to claim 1, characterized in that the screen is rotated by an angle of 2α. 3. The method according to claim 1, characterized in that the first and second screens are rotated in the same direction by an angle α.

Images