EP1019776A1

EP1019776A1 - Method and device for producing an image which can be represented in 3-d

Info

Publication number: EP1019776A1
Application number: EP98946376A
Authority: EP
Inventors: Helmut Würz
Original assignee: Individual
Current assignee: Individual
Priority date: 1997-08-20
Filing date: 1998-08-19
Publication date: 2000-07-19
Also published as: WO1999009449A1; DE19736158A1

Abstract

The invention relates to a method and a device for producing or generating an image which can be represented in 3-D. According to the invention, several complete pictures of an object being represented are taken from different laterally staggered directions, in a single procedure. The pictures are preferably taken through a series of cylindrical dispersing lenses which are arranged in the beam path in front of the camera objective. The resulting picture, e.g., a paper photograph or a transparency, is viewed through a series of cylindrical magnifying lenses side-by-side, the width of said magnifying lenses being equal to the width of the individual pictures. The inventive method is also suitable for CAD representations and for a 3-D television.

Description

Description:

Method and arrangement for producing a spatially reproducible image

The invention is based on an image that can be recognized by a viewer as a spatial image without additional aids.

There are a number of methods and devices with which images of objects can be produced, which can then be recorded as a spatial image by the viewer.

It is known that the impression of spatial depth can be created on a two-dimensional image display medium, for example a printed sheet of paper or a screen, by displaying on the image display medium a so-called parallax panorama diagram, which is then viewed through a lens grid arranged in front of the image plane . The parallax panoramagram is a representation of the three-dimensional object, which consists of several corresponding partial images nested in one another in the lenticular grid. The lenticular grid serves to optically separate the partial images from one another, so that in the

Viewing from a certain viewing direction, only the partial image belonging to this viewing direction can be seen. For this purpose, the lenticular grid is arranged in front of the panorama diagram in such a way that the focal plane of the lenses essentially coincides with the image plane. The light emanating from a point in the image plane is then refracted by the lenses into a substantially parallel beam which forms a certain angle with the normal to the image. The angle between the viewing direction and the image normal is, according to the laws of geometric optics, the distance between the point under consideration and the optical axis associated lens determined. The desired assignment between the partial images and the viewing direction can thus be achieved in that the partial image elements are arranged in suitable positions with respect to the lenticular grid. In practice, a lens grid consisting of vertically extending cylindrical lenses is usually used, and the viewing angle is varied horizontally, so that in particular the different parallaxes for the left and right eyes of the viewer are reproduced, thus giving the impression of spatial depth.

In a known method (DE-OS 23 48 700) an object is recorded sequentially with a camera. For this purpose, the camera must be moved between the individual shots.

Also known is an additional device (DE-OS 21 42 099) that continuously moves a serial camera in an arc around the subject. Here, too, several recordings are made with one camera.

In another known method (DE-PS 26 11 717) exposures are also carried out sequentially and a relative movement between the original and the light beam is carried out.

A method is known from EP 0 583 766 A1, in which the three-dimensional structure of the object is represented by a sequence of images which were recorded from a plurality of viewing directions varying around a central viewing direction. The image data is entered into a computer that calculates the parallax panoramagram, which is then printed on an image display medium. With the help of the computer, further interim caste images are synthesized that correspond to additional viewing directions.

In another patent DE 44 16 935 C2, a three-dimensional coordinate system (Cartesian), which is defined by a set of vectors, is used to calculate the panoramagram of an object, the viewer being at a specific location at a finite distance from the object and the gaze always remains on a certain view target point. The calculation here provides that the panoramagram contains several 2-D image elements assigned to each point of the three-dimensional object. This presupposes that sufficiently fine lenticular grids are used and the line elements are in the focal plane of the lenses.

In all previously known methods, the nested partial images or 2D image elements are located in the focal plane of the lenses and are essentially refracted as a parallel beam which forms a certain angle with the image normal. For this reason, it is necessary to use relatively fine lenticular grids in order to achieve adequate resolution.

In these methods, the resolution of the three-dimensional image is determined by the lenticular grid. The resolution of the image display medium used should be a multiple of this resolution, so that nested partial images with corresponding details can still be displayed under a lens. This leads e.g. B. with TV monitors already to major almost unsolvable problems. The The horizontal extent of a pixel on a 57-cra screen is approximately 0.6 mm. If one now assumes that for the construction of a parallax panoramagram with four partial images per lens, four image points are used for each partial image, the individual lens should have a width of 9.6 mm. Since the pixels are now in the focal plane of the lenses, the light is deflected by the monitor as a parallel beam in 4 different directions. As a result, the image is only 1/4 of the monitor resolution and can also have gaps in the form of dark stripes.

The invention has for its object to provide a simple method and an arrangement for producing a spatially reproducible image, which is small

Allow effort and even a coarse screening result in a streak-free undistorted view.

To achieve this object, the invention proposes a method with the features mentioned in claim 1.

The invention also proposes an arrangement with the features mentioned in claim 15.

Further developments of the invention are the subject of dependent claims.

The invention is based on the idea that the spatial object to be imaged produces several complete images which show the object from several laterally offset directions, each individual lens being assigned a complete image. The position of the image plane relative to the lenticular grid is not, as in the known methods, in the focal plane of the lenses, but, like in the case of a magnifying glass, between the focal point and the lens at a specific distance which is related to the depth and focal length.

An actually existing lenticular grid, which consists, for example, of several diverging lenses, is used to generate the images. However, this lenticular grid can also be simulated mathematically. The raster is imaged with the virtual images generated in the diverging lenses as an entire image by a camera, and this imaging can also be carried out arithmetically.

The areal complete image of the object generated corresponds to the individual views of the object taken from respectively adjacent starting points at a distance of the lens width of the individual lens of the lenticular screen. Each individual image has the parallax differences in all pixels or pixel sets in relation to its spatial depth compared to its neighboring images, which periodically repeat for each pixel or pixel set of the same depth point at a constant distance from image to image, as long as they can still be found in the respective image section is.

The following applies:

b * f a = distance from image to image a = b - b = lens width of the single lens

T + f of the lenticular screen f = focal length of the single lens of the lenticular screen T = depth of the virtual pixel in the three-dimensional spatial image

Basically, the recording of the object from two directions lying side by side is sufficient so that two images of the object are generated. However, the larger the number of individual images, the better the viewing possibility for the viewer, i.e. the greater the angle that enables spatial perception.

If the images lie between the lens and the focal point of the lens when viewed, the spatial image lies behind the lenticular grid. The effect achieved is comparable to a window through which you can look through.

If the images are arranged behind the focal point of the magnifying glasses, the spatial image appears to be in front of the lenticular grid. It gives the impression that the object is floating in the room.

In a further development of the invention it can be provided that the width of the magnifying glasses is equal to the width of the images. Then a magnifying glass is arranged at the appropriate distance before the individual image when viewing the images.

In a further development of the invention, it can be provided that the images are produced or generated in a shortened manner in the lateral direction. Then the resulting overall image retains approximately the same ratio of height and width, so that it can be more easily recognized by the viewer.

Images of the object from the different directions can be produced optically, in particular in further training, with the help of an optical image. However, it is also possible and is proposed by the invention that the images of the object are generated by calculation. This is particularly important or useful if the objects to be imaged are of a geometric type, that is to say essentially consist of lines. This type of generation of the images can be carried out, for example, if the objects are also objects which can be determined by calculation, that is to say, for example, in CAD programs. Since the imaging equations of optical systems are known, the images can also be easily calculated. It is also possible, for example, to carry out an optical imaging on an electrically scannable device, for example a CCD sensor, and then to have the effect of the lenses calculated. This is also possible with normal pictures, not only if the objects to be displayed consist of lines.

In a further development of the invention, it can be provided that the images are generated simultaneously. When producing the images optically, this means that only a single image is required for a still image.

In particular, it can be provided in the production of the images that these are produced with the aid of a series of adjacent, in particular cylindrical, dispersion lenses which are arranged in the beam path for the optical generation of the image. This is the simplest way to create a large number of images of the same object from laterally offset directions on a flat surface. In particular, it can be provided that the diverging lenses are arranged and designed in such a way that the images produced with their help lie next to one another in a plane with little or no lateral spacing.

According to the invention, the images can be generated, for example, on a photographic film, that is to say in a camera or film camera, or else on a monitor or a screen.

The invention also proposes an arrangement with the features mentioned in claim 15. The device for viewing the image can be part of the image, or an additional device that the viewer has with him in a manner similar to a magnifying glass or an enlarger for slides.

In particular, it can also be provided here that the width of the magnifying glasses is equal to the width of the images.

In a further development it can be provided that the recording device is designed such that the images are produced in a shortened manner in the lateral direction.

The recording device can have an optical camera or be formed by it. It is also possible that the recording device has a video camera if spatial processes are to be recorded and reproduced.

The recording device can also be designed such that it calculates the images mathematically. The mapping equations of optical devices such as lenses, mirrors or the like are known. It is therefore possible to also compute an image of an object. This is especially true if the objects are also in the form of equations or numerical values, such as in CAD programs.

According to the invention it can be provided that the recording device generates all images simultaneously or quasi simultaneously.

One way of generating several laterally shortened images is to produce the images using a series of cylindrical diverging lenses lying next to one another.

According to the invention, it can be provided that the images are generated and / or reproduced on a screen. This is also to be understood to mean that the images are first generated on an electrically scannable intermediate arrangement and only then are they generated on a screen with visible information.

Further features, details and advantages of the invention result from the patent claims, the wording of which is made by reference to the content of the description, the following description of a preferred embodiment of the invention and with reference to the drawing. Here show:

Figure 1 is a schematic representation of the creation of an image from several images.

FIG. 2 shows the arrangement for viewing the image generated in FIG. 1; Fig. 3 is a schematic illustration for explaining the composition of several images;

4 shows the beam path to explain the image;

5 shows the beam path with three magnifying glasses lying next to one another;

6 - 9 are schematic representations for further explanation.

1 shows, in a very simple overview, one possibility of how a series of adjacent images can be produced from different directions in an image plane 2 from a spatial object 1 with the aid of an optical system. To produce an image of the object 1, an optical system is provided, which is implemented by the lens 3 shown schematically. The lens 3 can be a photographic lens. The lens 3 would normally produce a two-dimensional image of the object 1 in the image plane 2. A device 4 for producing a plurality of individual images is arranged in front of the lens 3 in the beam path between the object 1 and the image plane 2. It is a series of cylindrical diffusion lenses 5 lying next to one another, that is to say, for example, a glass body which is provided with concave grooves on one of the two sides. An image of the complete object 1 falls on each of the cylindrical lenses formed thereby, all of the diverging lenses seeing the object 1 from a different direction. These different directions are laterally offset transversely to the cylinder axes of the lenses. Through the diverging lenses 5 a shortened image of the object 1 is generated in this direction, which with the help of Lens 3 is imaged in the image plane. This results in a plurality of images 6 of the object 1 lying side by side in the image plane 2. The device 4 is designed such that the individual images 6 of the object 1 are arranged next to one another with little or no spacing in the image plane 2.

A viewing device 7, which contains a large number of cylindrical magnifying glasses 8, is used to view the individual images 6 produced in this way, for example color photographs. The width of the magnifying glasses 8, in other words the grid of the magnifying glasses 8, corresponds exactly to the width of the individual images. As a result, a single magnifying glass 8 is arranged in front of each individual image, through which this image can be viewed. The distance between the

Image plane 2 and the viewing device 7 are selected such that the image plane 2 lies behind or in front of the focal plane of the magnifying glasses 8. This creates a virtual image that is enlarged when viewing the individual images.

The size of the grid, i.e. H. the width of the individual figures 6 and the cylindrical magnifying glasses 8 can be relatively large.

It is possible, for example, to design the viewing device 7 as part of the finished paper image, but also to design the viewing device 7 in the manner of an image viewing device for transparencies.

The creation of the spatial impression by viewing several individual images is now explained with reference to FIG. 3. 3 schematically shows two adjacent images in the form of an arrow 9, a cylindrical magnifying glass 8 being provided for each arrow 9. Since the images of the object, in this case the arrows 9a, 9b, are taken from different directions, they are also at different locations in relation to their magnifying glasses 8a, 8b. When looking at arrow 9a through magnifying glass 8a, an image b is formed. The following equation applies:

C (= correction factor) / f = b / (T + f)

Or in other words:

C = b * f / (T + f)

The sizes of the grid width b and the focal length f are constant in a given lens system, so that the correction contribution C depends directly on the depth T of the virtual pixel behind the lens grid. The tip of the arrow 9a is shifted upward by the amount C in the lower lens system in FIG. 3 compared to the upper system from the respective optical axes. The following then applies to the distance a of the same depth point from image to image:

a = b - C

Now to Fig. 4. The following equations apply to image size B and image width g:

B = G * f / (T + f)

g = B * T / G

When assembling the individual images into a spatial overall image, see FIG. 5, each individual lens system of the lenticular screen contains a reduced version of the when looking at the overall picture with the corresponding differences in direction in the illustration. The width of the respective image section corresponds to the width of the individual lens of the lenticular screen.

In the case of spherical lenses, an image is shown enlarged or reduced evenly, horizontally and vertically.

With a vertical cylindrical lens, the image is only enlarged or reduced horizontally.

It should be pointed out again that the position of the image plane in relation to the lenticular grid is not here in the focal plane of the lenses as in the other methods, but rather, as in the use of a magnifying glass, between the focal point and the lens at a specific distance from the depth and focal length is related.

As a result, the light rays of an image point are no longer emitted by a lens as a parallel beam, but rather as a diverging bundle, in the form that the image points of an identical spatial point converge under each lens in a rearward extension into a virtual spatial image point.

The described principle is based on the following principle:

All points in the image template that are to represent one and the same spatial virtual point must be placed in the image template in such a way that the rays converge in this spatial point after refraction, ie intersect at this point. This In principle, considerations apply to both spherical and cylindrical lens grids.

The essential and decisive advantage of the new method proposed by the invention compared to the previous method is that the fine lenticular grid required in the known methods and below the resolution of the eye is no longer necessary here. Grid widths can be selected, which may be relatively large, since even large individual images continuously merge into an overall image without a disturbing transition. The upper limit of the grid width depends on the viewing distance. At least two lenses of the lenticular screen with their partial images and the differences in direction have to be imaged on the retina of the eye in the area of sharp vision.

Example:

In a short form, a method for the calculation and graphic representation of 3D images is shown below using the previously developed equations. However, this method can also be used as the basis for the calculation and representation by appropriate computer programs. A cube serves as an example.

1.) Define grid width b of the playback grid, e.g. B. 7 mm

2.) Set the focal length f of the playback grid, e.g. B.

15 mm

3.) Define depth T of the room image, e.g. B. 30 mm for the front cube surface 4.) Calculate the image size B = G * f 12 * 15 = = 4

(T + f) (30 + 15)

When using cylindrical lenses for the reproduction grid, the image is only shortened horizontally.

5.) Calculation of correction factor b * f 7 * 15 c = = = 2.33

(T + f) (30 + 15)

6.) Calculate distance a of the same depth point from image to image a = b - c = 7 - 2.33 = 4.67

7.) Place the calculated image size at a distance a in the grid in the image plane. Image details that are outside the image section are no longer shown in this image.

Depth of the front cube surface 30 mm B = 4 mm c = 2.33 mm a = 4.67 mm

Repeat the calculation steps 3 to 7 described for the calculation of further depth points.

Depth of the rear cube surface 40 mm B = 3.27 now c = 1.9 mm a = 5.1 mm

8.) Connect all corner points of the front cube surface with the corner points of the rear cube surface by lines. The distance g of the image plane to the lenticular grid is calculated so that the average depth between the front and rear cube surface is used.

f * T

(T + f)

application areas

In addition to the already mentioned generation of paper images, which can be viewed with the aid of the magnifying glasses, television can also be improved with the aid of the described method. Recording with optically imaging components or instruments and storage on a photo layer or electronically via image converter can be carried out with the aid of the lens grid. A lens grid with diverging lenses is used for the picture, and the reduced virtual partial images of the object in the lens grid are imaged on the storage layer via the camera lens. In these partial images, the differences in direction, which correspond to the respective depths of the object, already exist. A completely rastered image template is thus obtained from individual images of an object, taken from neighboring starting points, with all information about the room, the brightness and color in a single process. If a lenticular grid with positive lenses in the correct size ratio is used for the reproduction and placed in front of the image plane, the recorded object appears as a three-dimensional spatial image. Such a lenticular grid can, for example, be attached directly to the television set, possibly also subsequently. Another possible application is the three-dimensional representation of CAD drawings on a screen. In this case, the individual partial images can be calculated using the known imaging equations and displayed in a grid on the screen, in front of which a front lens with the cylindrical magnifying glasses can then be attached in the correct grid.

Using the subsequent paraxial image controls (CAD drawings with an accuracy of 16 decimal places), a complete correspondence can be determined between the equations described at the beginning and the drawing method, which are necessary for the calculation of the images as a basis.

In the case of a theoretical diverging lens (FIG. 6) with the focal length f = 20 mm, an object with the size G = 20 mm (large arrow), which is at a depth of T = 60 mm from the main points of the lens, becomes a virtual image (small arrow) with the size B = 5 mm at a distance g = 15 mm from the main points of the lens.

In the case of a theoretical converging lens (FIG. 7) with the focal length f = 20 mm, an image with the size B = 5 mm (small arrow) at a distance g = 15 mm from the main points of the lens as a virtual image (large arrow) with the Size G = 20 mm at a depth of T = 60 mm from the main points of the lens.

From this fact it can be seen that if the virtual image created by an object through a diverging lens is viewed with a converging lens in the arrangement as a magnifying glass, the image can again be seen exactly as a spatial object. The following illustration uses the drawing method to check whether this knowledge can also be transferred to a lenticular grid with a total of 5 lenses.

8 shows a lenticular grid made of diverging lenses and a paraxial image construction, with the following theoretical lens data:

Grid width b = 22 mm focal length f = 22 mm

Values:

Item size G = 22 mm depth T = 88 mm

Image size B = 3.97 mm

Distance a = 17.6 mm

Correction factor c = 4.4 mm

Image distance to the grid g = 18.03 mm

The calculated values and the drawn values correspond completely.

FIG. 9 shows a lenticular grid made up of converging lenses in the arrangement of magnifying glasses, with the same theoretical lens data as above:

Grid width b = 22 mm focal length f = 22 mm

The paraxial image construction of the virtual spatial image was carried out with the values calculated and drawn above. Values:

Image size B = 3.97 mm

Distance a = 17.6 mm

Correction factor c = 4.4 mm image distance to the grid g = 18.03 mm

virtual room size G = 22 mm depth T = 88 mm

From this result it can be seen that with identical

Lens data for the recording grid and the playback grid, the real object and the virtual spatial image are completely identical in size and depth.

It gives the viewer the same spatial impression as the real spatial object and is reproduced 1: 1 as in this case. It takes into account the natural way of seeing the human anatomy (physiology) such as eye relief, viewing depth and convergence of the visual axes. Therefore, practically everyone can see the 3-D scene and feels no effort even when viewed for a long time. It is possible to focus the eyes on nearby and distant objects.

According to the invention, this consideration led to a lenticular screen with diverging lenses and the resulting advantageous possibility of obtaining finished partial images with all the parallel differences of the three-dimensional object in one go. Since these drawing files are only available virtually, the lenticular grid must be displayed as an entire picture. This is advantageously done with a commercially available camera, regardless of whether a photo, digital or video camera is used.

Claims

Claims:

1. A method for producing or generating a spatially reproducible image in which 1.1 several complete images (6) are generated from the object to be imaged (1)

1.2 show the object (1) from several laterally offset directions and

1.3 which are arranged in a plane (2) next to each other with a small lateral distance, and in which

1.4 in front of the illustrations (6) a series of magnifying glasses (8) are arranged next to each other

1.5 have a distance of more or less than the focal length from the illustrations (6) and 1.6 of which each magnifying glass (8) is assigned to an illustration (6).

2. The method of claim 1, wherein the width of the magnifying glasses (8) is equal to the width of the images (6).

3. The method of claim 1 or 2, wherein the images (6) are produced shortened in the lateral direction.

4. The method according to any one of the preceding claims, in which magnifying glasses (8) are used as magnifying glasses.

5. The method according to any one of the preceding claims, wherein the images (6) are generated optically.

6. The method according to any one of the preceding claims, wherein the images (6) are generated by calculation.

7. The method according to any one of the preceding claims, wherein the images (6) are generated simultaneously.

8. The method according to any one of the preceding claims, in which the images (6) are produced or produced with the aid of a series of scattering lenses (5) lying next to one another, which are in the beam path to the optical

Generation of the illustrations (6) are arranged and have a diverging effect in the lateral direction.

9. The method as claimed in claim 8, in which the diverging lenses (5) are arranged and designed in such a way that the images (6) produced with their help lie next to one another in a plane (2) with little or no spacing.

10. The method according to claim 8 or 9, wherein the diverging lenses (5) are cylindrical lenses.

11. The method according to claim 8 or 9, in which the diverging lenses (5) act in the direction transverse to their direction of diverging as converging lenses.

12. The method according to any one of the preceding claims, in which at least two rows of scattering lenses arranged one above the other and / or at least two rows of collecting lenses arranged one above the other are used for imaging.

13. The method according to any one of claims 8 to 12, in which the diverging lenses (5) are simulated by computer.

14. The method according to any one of the preceding claims, wherein the images (6) are generated and / or reproduced on a screen.

15. Arrangement for producing or generating a spatially reproducible image, with

15.1 an image recording device (4), the

15.2 records several images (6) of an object (1) from laterally offset directions in one plane (2), as well as with

15.3 an image viewing device (7), the

15.4 has a series of magnifying glasses (8) arranged next to one another, 15.5 each of which is assigned to one of the illustrations (6) and 15.6 is at a distance of less or more than the focal length from this.

16. The arrangement according to claim 15, wherein the width of the magnifying glasses (8) is equal to the width of the images (6).

17. Arrangement according to one of claims 15 to 16, wherein the magnifiers are cylindrical magnifiers (8).

18. Arrangement according to one of claims 15 to 17, wherein the recording device (14) is designed such that the images (6) are produced shortened in the lateral direction.

19. Arrangement according to one of claims 15 to 18, wherein the recording device comprises an optical camera.

20. Arrangement according to one of claims 15 to 19, wherein the recording device comprises a video camera.

21. Arrangement according to one of claims 15 to 20, wherein the recording device calculates the images (6) by calculation.

22. Arrangement according to one of claims 15 to 21, wherein the recording device generates all images (6) simultaneously.

23. Arrangement according to one of claims 15 to 22, in which the images are produced with the aid of a row of diverging lenses (5) which are arranged next to one another and are arranged in the beam path for producing the images (6).

24. The arrangement according to claim 23, wherein the diverging lenses (5) are cylindrical lenses.

25. Arrangement according to one of claims 15 to 24, in which the cylindrical diverging lenses (5) are arranged and designed in such a way that the images (6) generated with their help lie next to one another in a plane (2) with little or no distance.

26. Arrangement according to one of claims 15 to 25, in which the images (6) are generated and / or reproduced on a screen.

27. Arrangement according to one of claims 15 to 26, in which the object (1) is present as a computational model.