RU2350042C2 - Stereoscopic video imaging method and device - Google Patents

Abstract

FIELD: physics.

SUBSTANCE: stereoscopic video imaging device contains video cameras with converging optical axes, screens with polarisers in front, spectacles with analysers, beam-splitting plate through which screens are optically interconnected thus receiving effective transmitted and reflected light fluxes. Herewith the screens lying in vertical planes angularly are related to each other thus keeping symmetry of their central parts relative to the beam-splitting plate. The latter is inclined to the screen designed for receiving effective transmitted light flux, at an angle φ chosen from the range 0° to 90°, while screen inclination angle β is specified by ratio: β=2φ±α, where φ is angle of design screen inclination providing left and right images respectively.

EFFECT: improved quality of stereoscopic video imaging in stereoscopic tele and video systems, enabling shooting by video cameras with converging optical axes of objectives.

6 cl, 6 dwg

Description

The invention relates to methods and devices for producing stereoscopic television and video images (hereinafter referred to as stereoscopic images) and can be used in science, education, medicine, production, including microscopy, endoscopy, telemedicine, underwater television, where it is required to obtain real-time high-quality color 3D images objects.

In stereoscopic systems, the volume of the observed image is achieved by presenting the viewer with two flat images (stereo pairs) corresponding to the left and right angles of observation 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 devices for obtaining stereoscopic images presenting images of both angles in the common observation plane, when reproducing a stereo pair obtained by such a scheme, a distortion of the stereoscopic image arises, called a curvature of the screen plane [N. Valius. Stereoscopy. M .: Publishing house of Acad. USSR Science, 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.]. With this distortion, objects located in the center of the screen appear closer to the observer than objects at the edges of the screen [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.]. In this case, the device’s flat screen appears to be curved and the same thing happens with other planes throughout the reproduced space. As a result, the viewer receives a false impression about the spatial arrangement of objects in the observed scene and may make an erroneous decision.

A known method for 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 flux coming from the left and right images, placing a matrix of micropolarizers in front of the screen, and observe the image on the screen through glasses with analyzers. (US patent No. 6593959, IPC: H04N 15/00).

A device for implementing this method is known, comprising a liquid crystal display, a matrix of micropolarizers placed in front of the display screen, a device for processing a video signal, glasses with analyzers. The matrix of micropolarizers consists of alternating regions with first and second directions of linear polarization. The video processing device provides the display of fragments of images for the left and right eyes, in which fragments of the left image are displayed opposite the micropolarizers with the first direction of polarization, and fragments of the right image are opposite the micropolarizers with the second direction of polarization. (US patent No. 6593959, IPC: H04N 15/00).

These method and device are characterized by insufficient quality of the reproduced stereoscopic image due to a twofold decrease in resolution, as well as the curvature of the planes in the stereoscopic image when reproducing stereo pairs obtained by shooting with the converging optical axes of the camera lenses.

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. (US patent No. 6975345, IPC: H04N 13/04).

It is also known a device for implementing this method, comprising a display with a cathode ray tube, a liquid crystal modulator with the ability to switch between the first and second states of polarization, a modulator control device with the ability to synchronize the switching of the modulator with playback of the left and right images. (US patent No. 6975345, IPC: H04N 13/04).

These method and device allow to obtain a stereoscopic image without lowering the resolution, however, they impose increased requirements on the display, which should display frames with double frequency. In addition, when reproducing stereo pairs obtained by shooting with converging axes, the resulting stereoscopic image is characterized by insufficient quality due to the curvature of the planes in the reproduced space.

Closest to the claimed method is a method of obtaining stereoscopic images, in which using two video cameras receive images of the left and right angles of a three-dimensional scene, reproduce the images on two screens mounted at an angle of 90 ° to each other and optically connected using a translucent mirror , which is positioned at an angle of 45 ° 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. (Patent for invention US No. 2845618, NKI: 340-369).

Closest to the claimed device is a device for obtaining stereoscopic images, containing two cameras for receiving images of the left and right angles of a three-dimensional scene, a device for converting the images into a television signal, a transmitting device, a receiving device, two displays with cathode ray tubes, two polarizers , translucent mirror for combining images of display screens in the common observation plane and glasses with analyzers for separate binocular observations of combined images. Displays are placed symmetrically with respect to the translucent mirror at an angle of 90 ° to each other. (Patent for invention US No. 2845618, NKI: 340-369).

These method and device allow 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 double frame rate. However, when the cameras are oriented according to a scheme with converging axes, the observed stereoscopic image is characterized by insufficient quality due to the curvature of the planes in the reproduced three-dimensional space.

The objective of the invention is the development of a method and device for obtaining stereoscopic images that allow you to play stereo pairs from cameras with converging optical axes without the effect of curving the plane 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 cameras with converging optical axes, the image from which is reproduced on two screens mounted in vertical planes at an angle to each other and optically connected to each other using a beam splitter, while the screens are arranged with the symmetry of their central parts relative to the beam splitter plate, according to the proposed solution, determine the angle of the screens, for which place the plane the right rectangular lattice in the convergence plane of the cameras, reproduce the left and right images of the lattice on the corresponding screens, take the image of the lattice from each screen and after combining the images in the center, find at least two pairs of conjugate points of the left and right lattice images, determine the midpoint for each pair, according to the found midpoints, taking into account the position of the optical centers of the viewer's eyes, determine the estimated positions of the screens for the left and right eyes, determine the angle α between them, while m one of the screens is set in one of the calculated positions, and the angle β of the location of the second screen is determined from the ratio

β = 2φ ± α,

where φ is the angle between the screen installed in the calculated position and the beam splitter plate, selected from the interval from 0 to 90 °.

The conjugate points of the left and right images of the lattice are determined by the intersection of the vertical lines of the images of the left and right angles of the lattice with a horizontal line passing through the center of the lattice. The calculated positions of the screens with images for the left and right eyes are found at the intersection of the straight lines passing through the optical center of the left or right eye of the viewer and the midpoints, with concentric semicircles having centers coinciding with the center of the grating and passing through the points of the left or right image of the grating. In this case, the optical centers of the viewer's eyes are equidistant from the center of the combined image of the lattice.

By conjugate points, we mean points in two images of the same object, which are images of the same point of this object. In this case, we mean the images of the object obtained from the points of view corresponding to the viewing angles of the object with the left and right eyes (stereo pair) [Valius N.A. Stereoscopy. M .: Publishing house of Acad. Science of the USSR, 1962. 379 p.].

In known stereoscopic video 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 α between these planes, determined according to the proposed method, the quality of the obtained stereoscopic image increases, namely, the curvature of the planes in the stereoscopic image is eliminated.

The claimed technical solution proposes a method and device that allows you to observe a stereo pair obtained by shooting with converging axes, as a stereoscopic image with undistorted reproduction of planes.

The invention is illustrated by drawings, in which Fig. 1 shows a diagram for obtaining a combined image of a lattice in the claimed method, Fig. 2 shows a diagram for determining the design positions of screens in the claimed method; figure 3 presents an embodiment of a device implementing the inventive method; figure 4, 5 and 6 demonstrate the elimination of distortion in the image of the test rectangular lattice using the proposed method.

The positions in the drawings indicate: 1 - test flat lattice, 2 - convergence plane of video cameras, 3, 4 - video cameras, 5, 6 - images of the left and right angles of the flat lattice, 7, 8 - display screens, 9 - midpoint for a pair of paired points left and right images of the lattice, 10, 11 are the conjugate points of the images of the left and right angles of the lattice, 12, 13 are the optical centers of the eyes of the viewer, 14, 15 are the calculated positions of the screens with left and right images, respectively, 16 is a horizontal line passing through the center gratings, 17, 18 - polarizers, 19 — glasses with analyzers, 20 — beam splitter plate, 21 — volumetric object under investigation, 22 — viewer, 23 — screen image 7 observed by the viewer in the light reflected from the beam splitter, 24 — stereoscopic image observed.

According to the proposed method for reproducing images from cameras with converging optical axes, the angle of the image screens for the left and right eyes is first determined. For this (Fig. 1), a flat lattice 1 is placed in the convergence plane of 2 cameras 3 and 4, arranged according to a scheme with converging optical axes. An image 5 of the left angle view of the lattice received from the video camera 3 is displayed on screen 7, and screen 8 is an image 6 of the right angle view of the lattice received from the video camera 4. The resulting images of the lattice 5 and 6 are taken from the screens and aligned in the center. At least two pairs of conjugate points are found, for example, a pair of conjugate points 10 and 11 and any other pair. For each pair, a midpoint is determined, for example, midpoint 9 for a pair of conjugate points 10 and 11.

Based on the found midpoints, taking into account the positions of the optical centers of the eyes of the viewer 12 and 13 (FIG. 2), the calculated position 14 of the screen with the image intended for the left eye is determined. Similarly determine the estimated position 15 of the screen with the image intended for the right eye. Measure the angle α between the found calculated positions of the screens 14 and 15. Next, set one of the screens (first screen) in the calculated position, set the beam splitter plate at an angle φ to the first screen, arrange the center of the second screen symmetrically with respect to the plate with the center of the first screen, set the second screen - at an angle β to the first screen, and the angle β is determined from the relation

β = 2φ ± α (1)

The angle φ is selected from a range of values from 0 to 90 °, in a preferred embodiment of the invention close to a value of 45 °.

In this case, the conjugate points of the left and right images of the lattice, for example, points 10 and 11, are determined by the intersection of the vertical lines of the images of the left and right angles of the lattice with a horizontal line 16 passing through the center of the lattice.

The estimated position 14 of the screen with the image for the left eye (figure 2) is found at the intersection of the straight lines passing through the optical center 12 of the left eye of the viewer and the found midpoints, for example point 9, with concentric semicircles having centers coinciding with the center of the grating and passing through the points of the left image 5 of the lattice, for example, point 10. Similarly, the calculated position 15 of the screen with the image for the right eye is found at the intersection of the lines passing through the optical center 13 of the right eye of the viewer and the found average t hibernation, for example 9 point, with concentric semicircles having centers coinciding with the center of the grating, and passing through the right image lattice point 6, for example a point 11. Thus the optical centers of the viewer's eyes are equidistant from the center of the combined image of the lattice.

The device (Fig. 3) for implementing the proposed method comprises video cameras 3 and 4 arranged so that the optical axis of their lenses converge at one point, screens 7 and 8 with polarizers 17 and 18 installed in front of them, glasses with analyzers 19, a beam splitter plate 20 through which the screens are optically connected with the possibility of obtaining useful transmitted and reflected light fluxes, while the screens are located in vertical planes at an angle β to each other to ensure symmetry of their central parts relative to the beam splitter nooy plate.

In this case, the beam splitting plate is located relative to the screen 8, intended to obtain a useful transmitted luminous flux, at an angle φ selected from the interval of values from 0 to 90 °, and the screen angle β is determined from relation (1).

The polarizers 17 and 18 are oriented with the possibility of polarizing the light flux in orthogonal directions, and the analyzers in the glasses 19 are oriented accordingly. As a beam splitting plate 20, a translucent mirror, for example, with a multilayer dielectric coating, is used.

The device (figure 3) works as follows. The image of the left angle of the investigated volumetric object 21 is obtained using the video camera 3 and displayed on the screen 7. The image of the right angle of the object 21 is obtained using the video camera 4 and displayed on the screen 8. Spectator 22 observes the screen 8 through the glasses 19 in the light transmitted through the plate 20 and the image 23 of the screen 7 in the light reflected from the plate 20. In this case, due to the presence of polarizers 17, 18, the left eye of the viewer sees only the image of the screen 23, and the right eye - only the screen 8, as a result of which the viewer perceives the represented object as three-dimensional, that is, sees a stereoscopic image of the object. Since the angle β was chosen in accordance with relation (1), the viewer sees the screen 8 and the image of the screen 23 located in different planes intersecting at the desired angle α. As a result, in the flat images visible to each of the viewer's eyes, the geometric distortions that occurred during shooting due to the imbalance of the sensors of the cameras are compensated, and as a result, the viewer observes a stereoscopic image of the object without curving the planes.

Figure 4-6 shows the elimination of distortion in the image of the test rectangular grid located in the convergence plane of the cameras with converging axes, using the proposed method. Line 16 of Fig. 4 shows the positions of the conjugate points of the left and right images of the lattice corresponding to a stereo pair of the lattice obtained using two cameras with converging axes with a shooting base of 8 cm and a convergence distance of 30 cm, the lattice consisting of squares with a side of 4 cm.

Figure 4 shows that when presenting the left and right images of the investigated lattice in a common plane, which corresponds to known stereoscopic video systems, its stereoscopic image is curved. Figure 5 shows the definition of the estimated positions of the screens with left and right images of the lattice according to the claimed method. Figure 6 shows the elimination of the distortion of the stereoscopic image of the lattice that occurs after placing each of the screens in the corresponding design position, determined according to the claimed method.

4, the positions of the points of the observed stereoscopic image 24 are found at the intersection of visual lines drawn between the optical centers of the eyes of the viewer and the corresponding points of the left and right images. The stereoscopic image 24 (figure 4), obtained in a known manner of representing a stereo pair in one plane, is curved and does not correspond to the shape of a real object.

Figure 5 shows the location of the estimated position 14 for the screen with the left image and the estimated position 15 for the screen with the right image in accordance with the claimed method.

In Fig.6, the image 23 of the screen with the left image and the screen 8 with the right image are placed in the found design positions 14 and 15 and a stereoscopic image of the lattice 24 is built, and the image 24 is constructed according to the same rules as in Fig.4. It is noticeable that now the distortion of the lattice image is eliminated, and the shape of the observed image corresponds to the shape of a real object. Thus, the test stereo pair observed in known stereoscopic video systems as a stereoscopic image with a curvature of the screen plane can be reproduced without distortion using the proposed invention.

Claims (6)

1. A method of obtaining stereoscopic video images from two cameras with converging optical axes, the image of which is reproduced on two screens mounted in vertical planes at an angle to each other and optically connected to each other using a beam splitter plate, while the screens are arranged to ensure symmetry of their central parts relative to the beam splitting plate, characterized in that they determine the angle of the screens, for this place a flat rectangular lattice in the plane of the conv camera segments reproduce the left and right lattice images on the respective screens, take the lattice image from each screen and after combining the images in the center, at least two pairs of conjugate points of the left and right lattice images are found, the midpoint for each pair is determined from the found midpoints, taking into account the position of the optical centers of the viewer's eyes, determine the estimated positions of the screens for the left and right eyes, determine the angle α between them, while one of the screens is installed in one of design positions, and the angle β of the location of the second screen is determined from the ratio
β = 2φ ± α,
where φ is the angle between the screen installed in the calculated position and the beam splitter plate, selected from the interval from 0 to 90 °. 2. The method according to claim 1, characterized in that the conjugate points of the left and right lattice images are determined by the intersection of the vertical lines of the images of the left and right angles of the lattice with a horizontal line passing through the center of the lattice. 3. The method according to claim 1, characterized in that the calculated positions of the screens with images for the left and right eyes are found at the intersection of straight lines passing through the optical center of the left or right eye of the viewer and midpoints with concentric semicircles having centers coinciding with the center lattice, and passing through the points of the left or right image of the lattice. 4. The method according to claim 1, characterized in that when determining the estimated positions of the screens, the optical centers of the viewer's eyes are equidistant from the center of the combined image of the grating. 5. A device for producing stereoscopic video images containing video cameras with converging optical axes, screens with polarizers mounted in front of them, glasses with analyzers, a beam splitter plate through which screens are optically coupled with the possibility of obtaining useful transmitted and reflected light fluxes, while the screens are arranged in vertical planes at an angle to each other with the symmetry of their Central parts relative to the beam splitter plate, characterized in that the beam splitter Naya plate is positioned relative to the screen intended for receiving the useful transmitted light flux angle φ, selected from the range of values from 0 to 90 °, and the angle of the screens is determined from the ratio β
β = 2φ ± α,
where φ is the angle between the calculated positions of the screens with left and right images, respectively. 6. The device according to claim 5, characterized in that the screens are screens of displays with cathode ray tubes or plasma displays or displays on organic light emitting diodes (OLED).

Images