WO2011029440A1 - Image capturing and visualization providing a spatial image impression - Google Patents

Abstract

An apparatus and a method for photographing and reproducing images providing a spacial impression are disclosed. The reproduction of spatial photographs, e.g. stereophotographs, usually requires technical aids, such as eyeglasses or special monitors having restricted fields of view, placed close to the head. The method used aims to do away with such restrictions. To achieve this aim, several photographic capturing devices arranged vertically above each other are used, a virtual, periodic slow and soft vertical camera motion is calculated from the images thus obtained, and said images are reproduced on a commercially available monitor. The spatial impression is created as a result of the similarity of the motion thus simulated to the vertical motion of the human eye during walking. The method and the apparatus are suitable for 3-D representation on any devices that are suitable for reproducing image sequences, such as computer monitors, cinema screens, and electronic billboards. The vertical animation is not limited to the visualization of real images but can also be produced by rendering appropriate virtual camera positions of any 3-D models.

Description

Image acquisition and visualization with spatial image impression

description

 For documentation purposes and for aesthetic reasons, methods for spatial image acquisition and reproduction have been studied since the 19th century. The advantages of these methods in photographic applications, in contrast to traditional one-eyed photography, are that information about the 3D structures of the scene is also documented and that disturbing influences in the two-dimensional image such as "out of the heads lampposts", "Falling lines" or Reflections on visual elements can be separated from the viewer by the actual object and therefore less disturbing. Typical examples of 3D photography are stereo techniques in which two cameras are mounted horizontally and triggered simultaneously. Over time, many methods have been developed to deliver the two images separately to the eye of the beholder. The reproduction of the image material is associated with great effort. Thus, the observer wears either a stereo goggle with color or polarization filters or his location is - within autostereoscopic procedures - set within narrow ranges. In addition, about 5% of the population is unable to see stereoscopically.

[2] These disadvantages are avoided by the method described in US Pat. No. 4,429,328, which was developed for moving pictures and published under the brand VISIDEP. In this case, the images are switched by two vertically superimposed cameras with high frequency and so alternately supplied to the screen of the viewer. The human visual system calculates the three-dimensional structure of the recorded scene from the different amplitude of the movements in the image. However, the fast, abrupt change between the individual shots disturbs the viewing considerably. Sources even report discomfort as a result of prolonged consideration of such tiger "Wackelbilder." For example, in Internet wikis there is a comment from a former student involved in the development of "VISIDEP was a guaranteed headache generator" in 2008. The cited protection document indicates that the vertical arrangement of the pick-up positions produces less negative effects than horizontal. As a vertical distance of the recording points are assumed there 1-1, 5cm, as a frame rate of 4-lOHz. The reasons for the better results with vertical rather than horizontal camera arrangement were previously unknown. This is explicitly stated in the protective paper.

 [3] Later, the method for artificial computer-generated scenes was investigated (T. Barham et al., Some variations on VISIDEP using computer graphics, Optical Engineering 29 (12), 1990). By inserting intermediate images, the jumping process and its negative side effects are reduced. However, since the high frequency of movement is deliberately maintained by increasing the frame rate accordingly, the disturbing impression of a high frequency of movement and a jerky impression of movement remain in the reversal points. In particular, the high vertical frequency of movement makes it impossible for the observer to focus on the rapidly moving image sections in the vicinity without irritation; a contemplative preoccupation with the image content, as practiced in artistically demanding photographs, is therefore impossible.

The described invention has for its object to find a device and a method with which a spatial visual impression can also be achieved without aids and without restriction to the location of the observer, but in which no irritations arise.

[5] The object of the invention is based on the recognition of why a vertical camera arrangement as described in US Pat. No. 4,429,328 is used for the representation of spatial alternating pictures is more suitable than a horizontal one. The solution uses these findings for a device and a method for particularly advantageous recording, design, processing and presentation of the image material.

[6] This new finding about the particular suitability of a vertical camera arrangement is based on the fact that a vertically moving eye point belongs to the natural life experience of humans. The vertical component of its eye movement while walking can be modeled by a sinusoidal vertical movement at a frequency corresponding to its current pace. The image of stationary objects in its surroundings therefore moves with a distance-dependent vertical oscillation on the retina. Humans estimate the relative distances of elements of the scene based on this different vertical velocity. At a typical vertical amplitude of 2.5 cm and a step frequency of 1.5 Hz, the spatial separation of objects staggered in depth is comparable with the performance of the fully trained binocular stereo view, with a known measurement sensitivity of about 1 arc minute / s. in which a sensitivity of 20 arc seconds at 6.5 cm base are known as typical values ^'' wide. the distance estimation due to the vertical movement when walking in addition, while the man in the binocular stereo vision has the advantage that it acts in the entire field of view simultaneously. only one Restricted distance range around that of the current fixation point can analyze, it appears the rest as double images and is suppressed in further perception steps in the processing, the vertical movement of the eye induced movement of the object images on the retina in the entire scene perceived z spatial scenes represented by vertical motion appear more natural than conventional stereo images. [7] The solution of the problem according to the invention is realized by the following devices or methods.

[8] The choice of a movement frequency below 2Hz typically 0.5Hz.

 [9] An imaging device with a vertical arrangement of two or more digital sub-cameras. For the applicability by classical users of photographic equipment is particularly important that the device in the operation of a classic camera corresponds or can be combined with such.

 [10] The generation of the image sequence of a soft, e.g. Sinusoidal movement from these camera images by a suitable choice of the vertical distances of the sub-cameras or the assumed vertical position in the interpolation of intermediate images.

In the context of this document, the term "partial camera" means a combination of optics and a digital image sensor The device according to the invention combines two or more of these vertically stacked sub-cameras into a "3D camera".

The invention will be explained in detail with reference to embodiments and with the aid of figures.

[13] For a better explanation, the optics of the respective uppermost subcamera in all relevant figures is designated (101).

Figure 1 shows a device which is mounted instead of an interchangeable lens to a single-lens reflex camera with a large image sensor. It realizes the function of the described 3D by means of vertically superimposed optics. Camera. The optics depict the scene on the large image sensor in the camera body. Since these sensors are usually installed in landscape format, the camera must be held upright for shooting with the special lens drawn. For this purpose, for many SLR systems standard vertical handles - shaded here, which allow a comfortable hold and release in this position.

Figure 2 shows an embodiment of the 3D camera, which is plugged as an accessory in the hot shoe of a conventional camera.

FIG. 3 shows an embodiment of the 3D camera as an additional device to a flash unit.

FIG. 4 shows an embodiment of the 3D camera as an additional device in a camera tripod.

FIG. 5 shows an embodiment of the 3D camera, in which groups of extremely miniaturized wafer-level cameras are used as subcameras which are separated for the entire visible light spectrum (106) ^" or for the subregions red (105), green The same vertical array of sub-cameras is located a second time to the right of the first, with a different focal length from that of the first row (110) adjustable in pairs.

FIG. 6 shows an embodiment of the 3D camera as an additional device in an otherwise commercially available mobile radio device. FIG. 7 shows the 3D camera plugged together from independent partial camera components.

FIG. 8 shows a different arrangement of independent partial camera components from FIG. 7, in which the base width is realized by shifting the partial cameras relative to one another. By an exemplified construction of gear (102) and racks (103) ensures that the distance of the sub-cameras changes to the same extent.

[22] The inventive solution of the problem by a device according to the independent claims 1 and 3 or the claims derived therefrom and a method according to the independent claim 36 and claims derived therefrom and a reproducing apparatus according to claim 54 is achieved by a vertical arrangement of several sub-cameras , In the simplest case, the soft, sinusoidal impression of movement is produced by the fact that the recorded images of the sub-cameras are at the appropriate vertical positions within the desired sine oscillation and are reproduced sequentially and cyclically. With a simulated step frequency of 1.5 Hz and a frame rate of 15 Hz necessary for a smooth motion perception, the use of 5 sub-cameras is required. This results because the image sequence for a period must consist of 10 individual images (15 Hz frame rate / 1.5 Hz step rate). This number of frames can be generated by K = 5 sub-cameras in the following geometric arrangement and order. Since the usual display media only allow a constant frame rate, the vertical positions of the cameras are chosen so that the desired soft sinusoidal movement results. If the subcameras are numbered from top to bottom with k ₀ , ki..k ₄ , the subcamera k _{n has} a central one Symmetry plane the vertical distance

d _n = b (cos (pi * (n + 0.5) / K)) / 2

 [23] b here denotes the vertical amplitude of the simulated eye movement. For simulated walking movements, it can be assumed to be about 1.25 cm, resulting in a distance between the vertices of 2.5 cm.

[24] This value gives a distance di = -d2 = 7.3mm or d ₀ = -d ₄ = 11.9mm from the central vertical line of symmetry. The image sequence in the simulation of the simulated eye movement is k ₀ , k ₀ , ki, k ₂ , k ₃ , k ₄ , k ₄ , k ₃ , k ₂ , ki

 In this choice of b, however, the sub-cameras 0 and 1 or 3 and 4 have a distance of only 4.5mm. Therefore, in this arrangement, the cameras 0 and 4 are required in an embodiment with particularly small dimensions. Modern wafer-level camera modules have an edge length of less than 3 mm and are therefore particularly suitable for such a camera arrangement.

 In accordance with the invention, the non-equidistant arrangement of the sub-cameras described in this way achieves a pleasant, smooth motion representation in the generated image sequence compared to the prior art.

 At a later stage of the present document methods are described which can achieve the desired motion impression by a post-processing of the image material with the use of less than the five sub-cameras described here during recording. Therefore, the variants of the design of the cameras described below can be used for the task, even if you produce less than five - in an individual case, only two - vertically superimposed single shots.

[28] The so-called "dollhouse effect" is known from classical stereo photography, which results from the fact that the images for the left and right eye have not a typical human eye distance but a base width - sometimes many times greater leads to the seeks to enhance the spatial impression, as if the observer looks at a scene reduced by the corresponding factor by a factor shortened by the same factor.

 Such a reinforcement of the spatial impression is also achievable in the 3D camera described here by the distance between the sub-cameras is increased by a corresponding factor. According to the invention, the 3D camera can be designed so that this distance is variably adjustable, so that the photographer can determine the strength of the spatial visual impression by selecting the base width (which in the present vertical arrangement would actually have to be called "base height") during the recording Such a adjustability is particularly advantageously realized in such a way that the ratio of the spacings of the sub-cameras remains constant.

 This is the example of the later explained recording by a large number of miniature cameras in Figure 5 by two with a fixed slope or speed ratio coupled spindle drives (109) and (110), of which (109) the distance of the Partial cameras 0 and 4 relative to the central camera 2 increases or decreases and (110) the cameras 1 and 3 shifts by a proportionally smaller proportion. If the pitch of the spindle drives is the same, the speed ratio thereof must correspond exactly to the ratio between the distances of the sub-camera groups moved by them to the plane of symmetry.

[31] Another possibility that does not require moving sub-cameras is to provide the photographer with only a small number of discrete base widths by switching between multiple sets of sub-cameras before shooting or after shooting the pictures of the matching subcameras can be selected. Thus, for example, the five sub-cameras described by another sub-camera pair k ₅ and k ₆ at a distance of d ₅ = -d ₆ = do * d ₀ / di be added to the central vertical line of symmetry.

[32] The image sequence K ₅ , k ₅ , k ₀ , k ₂ , k ₄ , k ₆ , k ₆ , k ₄ , k ₂ corresponds geometrically exactly to the original image sequence with a base width increased by 62%. The addition of a further subcamera pair at a distance of d5 * d5 / d0 allows recording with a further 62% larger base width. In such an arrangement, each half of the sub-cameras remains unused for later playback. This can be avoided by the fact that the sub-cameras are mounted in a housing pluggable. The photographer can thus change the sub-cameras at positions kl and k3 to positions k5 and k6 and analogously those at positions k0 and k4 to positions k7 and k8.

 [33] When choosing the base width, the photographer will generally be guided by the idea of producing the strongest spatial visual impression in which the oscillating motion in the image is not perceived as disturbing. Experience has shown that this is certainly the case for motion amplitude up to 1% of the image height, but in some cases 5% of the image height is still tolerable.

 When using the focal length of a normal lens, this will reach the limit of 1% with 2.5cm distance of the top and bottom camera for objects at 5m distance, i. The closest object, including the ground, should be 5m or more apart. If, for example, the minimum distance to the imaged objects is instead 10m, the basic width can be doubled to 5cm according to the described criterion according to these criteria for a normal lens.

[35] An extreme case is an oblique aerial photograph with a minimum distance of the objects of 200m distance, where the optimal 3D camera according to the invention will have a basic width of typically approx. 100 cm. If large base widths are to be covered by the camera, but the housing should still be made compact, it is advantageous to run the camera so that the housing or a portion thereof vertically by 180 °, for example, as in the mobile device StarTac des Manufacturer Motorola can be opened. The sub-cameras are in this case in place of the display and keyboard and have so in the unfolded state, a significantly greater maximum distance than the height of the camera body corresponds. In addition, they are protected during transport. A similar effect can be achieved if the partial lenses are mounted on two housing parts, which are telescoped into each other and are pushed apart for recording.

[37] When designing the 3D camera, it makes sense that the ergonomics - apart from the additional setting option described for the basic width - deviate as little as possible from the operation of a conventional camera. It is therefore advantageously designed so that it has a trigger on the top and right or left of the optics areas for gripping and guiding with the hands.

[38] Another ergonomically particularly advantageous design of the 3D camera is to execute it as an additional component for a conventional camera. This can be done, for example, that the 3D camera can be plugged into the hot shoe of the conventional camera - as shown in Figure 2.

[39] The electrical connection between conventional camera and 3D camera transmits the triggering impulse. Other recording parameters for the conventional camera, such as exposure information, may also be transmitted and affect the exposure settings of the sub-cameras. In the opposite direction, the image information of the sub-cameras be transferred to the image memory of the conventional camera. A complete or partial transmission of information via radio-based communication, such as Bluetooth or WLAN is another possible implementation.

 [40] Using the hot shoe to connect the SD camera to a conventional camera has the disadvantage that the hot shoe itself can not be used to connect a flash. In a particularly advantageous embodiment, therefore, the 3D camera is integrated into a flash unit such that the housing of the flash unit picks up the sub-cameras - see FIG. 3.

[41] Instead of the hot shoe, the 3D camera can of course also be connected via other connectivity on demand with a conventional camera. Thus, it is possible, as shown in Figure 4, to accommodate the 3D camera in the top telescope segment of a conventional tripod, so that the photographer can use the 3D function as soon as the conventional camera has connected to the tripod.

[42] The combination of the 3D camera with a conventional camera allows the photographer to simultaneously make a 3D version and a conventional shot and use it as needed. For example, in the classical image presentation on paper - where he will use the conventional recording - and in the presentation on a dynamic medium, a TFT picture frame or the website, where he can use the 3D version.

[43] Special recording conditions may mean that a 3D camera - for example due to a high required photosensitivity - is not usable or - if the recorded scene has no depth graduation - has no effect. Even in these cases, the photographer will resort to conventional recording. [44] Even when using conventional 2D capturing, the sub-images of the 3D camera still have a value: when conventional camera and the 3D sub-cameras are calibrated to each other, the 3D sub-images can be evaluated for the overwhelming majority of the pixels by means of stereo evaluation the conventional recording to calculate the actual distance. The "depth image" obtained in this way can be used as a mask for the further image processing of the conventional recording, for example for the release of objects or persons or for the subsequent generation of distance-dependent blurring.

 [45] 3D cameras can also be realized in the form of a building block that can be configured by the photographer himself from subcameras. Thus, until 2005 under the name SL300r a digital camera of the company Kyocera was only 18mm high, in which the front lens of the optics is located on a narrow side. Such a camera can be extended to a sub-camera module when it is provided with a mechanical device for vertically interconnecting a plurality of sub-cameras and an electronic communication means for common setting of recording parameters and a common triggering and recording to obtain a 3D camera according to the invention. A possible mechanical connection via a dovetail construction is shown in FIG. 7. With mechanically compatible intermediate pieces which can be inserted between the stacked cameras by the photographer, the base width is variably adjustable.

A more advantageous geometric arrangement of flat modular sub-cameras results when the front lens of the optics is not on a narrow side, but as close to an edge on the adjacent largest surface of the camera body. If the mechanical stacking device is designed so that the longitudinal displacement between the cameras is variable adjustable, so that a variable base width without additional intermediate pieces is possible. It is beneficial to assist the photographer in using three or more of the sub-camera modules in the adjustment of the base width of the distance from each two adjacent modules is the same.

 [47] In the simplest case, this can be realized by concrete detent points for discrete shifts, so that it is easier for the photographer to set the same discrete shift each. Even easier to handle is a mechanical coupling, e.g. about gears and racks, as shown in Figure 8 for a sub-camera example.

Even a synchronized by the respective sub-cameras shift over an electromechanical drive provides the photographer a particularly convenient way of uniform adjustment of a variable base width available.

[48] Mobile phones are increasingly being equipped with digital cameras. If the corresponding optics are placed on the narrow side, a similar modular arrangement can be achieved. However, it is likewise possible to configure a 3D camera according to the invention as an additional function of a single mobile device by integrating several of these in the device instead of the commercially available individual camera - see FIG. 6.

[49] In the embodiments described so far, the sub-cameras are each mechanically independent units of optics and sensor. However, it is also possible to mechanically combine individual elements of the cameras. For example, the shutter for all sub-cameras can consist of a single mechanical unit as in the Nimstec analog Nimslo camera from the 1980s. The shutter is described on page 48 in ISBN 0-939617-00-5: Reel 3-D Enterprises' Guide to the Nimslo 3D Camera, 1988.

[50] Another possibility of mechanical combination consists of using the same sensor chip for more than one sub-camera. In one possible form they exist Subcameras eg only from separate optics, each illuminate different sections of one and the same large-scale image sensor. In this way, the 3D camera via interchangeable lenses, lens attachments or replaceable front optics for conventional digital cameras can be realized. The interchangeable lens carries here, for example, the optics of each sub-cameras or the lens attachment corresponding mirror or lens designs. Figure 1 shows such an arrangement.

 [51] In order to make the best possible use of the sensor surface, the images of the partial objectives must be arranged along the longitudinal side of the sensor chip. This requires either a portrait format. Upright configuration of the underlying camera body or handle options to comfortably hold the case in portrait orientation. Such portrait grips are commercially available as system accessories for SLR cameras. In conventional stereo photography, for all three enumerated embodiments in the 1980's, conventional stereo photography with horizontal arrangement of the photographs - single optics, front optics or face prism, e.g. under the brand name Stereotar or Steritar of the manufacturer Zeiss Ikon.

 [52] For the present task, however, they are not useful in the commercial form. Rather, their design parameters must be modified so that instead of the usually provided 6-8cm "eye relief" the optimum for the range between 5m and 10m base width of 2.5cm is realized special solutions with correspondingly reduced base width already exist under variants of Steritare however, intended and usable exclusively for macro shots in the short distance range in the sense of an inverse "dollhouse effect".

[53] Furthermore, it is advantageous to generate a larger number than the two images generated in conventional stereo photography. If the long side of the image sensor the Exceeds the sum of base width and narrow side of a field, the integration of further partial lenses is sufficient as shown in Figure 1 for the case of 3 fields. If the image sensor is smaller, the mentioned mirror or prism optics in the base width can be reduced. By pulling apart the two mirrors or prisms closest to the lens, it is additionally possible to produce a third partial image in that the objective additionally images the preceding scene without beam deflection by means of the resulting intermediate space.

 Another way to share optical elements for sub-cameras is the use of multiple overlapping aperture, the object side use a common and sensor-side individual optics. This is described for the case of three optics in US 20080204900.

 [55] A particularly simple solution to the problem arises if no movement occurs in the scene to be photographed or a periodic movement of scene elements in addition to the desired periodic movement of the camera height does not bother. In such a case, a conventional lens design can be used, in which the entrance pupil is about 1/3 larger than the desired amplitude of movement with fully open aperture. If one replaces the aperture of the lens by a height-shiftable aperture at most 1/3 of the maximum aperture and shifts the aperture with the desired low-frequency soft, periodic motion pattern, the lens generates in the image plane exactly the desired temporal image sequence, which was recorded by the sensor and can be recorded subsequently.

[56] A further solution of the stated problem, in which only a single lens is sufficient for generating discrete frames staggered in the eye-point height, consists in the application of the plenoptic principle as described in "Single Lens Stereo with a Plenoptic Camera", by Adelson and Wang , IEEE Transactions on Pattern Analysis and Machine Intel- ligence, vol. 14, no. 2, pp. 99-106, Feb. 1992. In turn, a standard objective lens is required, the entrance pupil of which at the greatest aperture clearly exceeds the amplitude of the vertical movement to be generated. The function-determining element is an image sensor which sits upright in the imaging plane of the optics with cylindrical lenticular lenses arranged horizontally on the surface, each of which distributes the light rays arriving from the objective onto at least n individual lines (where n is the number of desired partial images of the different vertical acquisition positions within the entrance pupil). To produce a vertically and horizontally homogeneous image sharpness, a change of the beam path that goes beyond the mentioned publication is required. The aperture must produce the same blur in the horizontal direction, which also receives a partial image in the vertical direction. This can be achieved, for example, by supplementing or replacing the standard diaphragm of the objective with a rectangular diaphragm which has the full aperture of the standard diaphragm in its height and which is reduced in width by a factor of n compared to the other. It is particularly advantageous for the universal use of the camera if the photographer can insert or remove the lenticular array and the rectangular aperture as needed and so reconfigure a standard camera with a large aperture in a short time in a 3D camera according to the invention. However, since the distance of the lenticular array from the image sensor is only a few microns, such an exchange is very susceptible to interference - such as dust. This hazard can be avoided by having a relay optic create a real image in an intermediate plane and the photographer can mount the lenticular a few microns away from this real image, or by designing the camera so that the entire conventional sensor passes through one second sensor is replaced with applied lenticular. As on the sensor with lenticular the picture multiple times only slightly different perspective is generated, the achievable image quality is much less sensitive to typical sensor errors such as hot pixels, etc. So it can be installed with the lenticular grid with more false pixels and therefore cheaper sensor, as for conventional photographic purposes. It is also advantageous to use for the detection of colors on the image sensor other than the usual Bayer pattern, for example, vertical, continuous, one pixel wide stripes alternately in the colors red, green, blue or red, green, blue, Green. Such an arrangement increases the resolution in the vertical direction and thus also the determinability of the vertical movement that is important for the impression. The increase in resolution in the vertical direction comes at the expense of horizontal resolution and thus leads to a more balanced overall impression, since the vertical resolution is reduced by the lenticular, which splits the image into interlaced fields, already compared to the horizontal resolution.

 [57] The above-described simple method for generating and processing the image material is based on the fact that the desired smooth motion sequence is realized in a sequence by a suitable arrangement of sub-cameras with predetermined non-equidistant distances and the direct reproduction of the images taken by these sub-cameras , For design, manufacture and use of a 3D camera, however, it is more advantageous if the vertical distances of the sub-cameras need not be set in advance to the soft sinusoidal profile of the image sequence to be generated, but are freely selectable. Two other variants of the method which ensure this can be found in the following descriptions.

The second variant of the method is to select the refresh rate per se much higher than the presentation time of the fields. This allows the Images k ₀ , ..., k _n -i of n sub-cameras, which are arranged in a freely selectable, eg equidistant vertical distance d ₀ , ..., d _n from the plane of symmetry of the vertical sinusoidal movement by repetitions for a different duration can be presented. A soft sinusoidal motion is now approximated by the fact that for the periodic image sequence b ₀ , i, b _m -x to be generated, the one most suitable partial image k _{j is} inserted at the position i, ie for the

[59] I k _j -b (cos (2pi * (j + 0, 5) / m)) / 2 I becomes minimal. The refresh rate is chosen so that m is greater than 2 * n.

The third and last variant of the method consists in the fact that not only the images taken directly by the sub-cameras are used in the reproduction of the sequence, but intermediate images are interpolated. The vertically superimposed cameras in number and position are much less strictly defined than in the two variants described so far. Methods and methods for image interpolation are known from various applications. The simplest solutions are found in commercially available televisions with "100 Hz technology" in which intermediate images are determined by interpolation.Such image interpolation methods are typically based on that in a first

Step, the optical flow of the image sequence is calculated, the intermediate images are formed by the fact that the flow vectors are completed by interpolation and smoothing and proportionally reduced and the respective interpolated intermediate image is reconstructed from the thus reduced flow vectors. As a result, the objects in the image have traveled only a proportionally reduced distance in the interpolated intermediate images. These known methods for flux calculation allow object movements in arbitrary directions on the two-dimensional image plane. However, since the images of the sub-cameras are vertically arranged stereo image couples, the possible direction of movement of the objects between the sub-images is limited to the respective epipolar lines. Therefore, instead of the described two-dimensional calculation of the optical flow, known methods for calculating intermediate views from stereo image pairs are better suited for interpolating the partial images, as are known, for example, in the case of a horizontal camera arrangement from the video conference. Since, in contrast to video conferencing, the real-time capability of the calculation is not important for the task in the photo, it is possible to place additional value on the image quality when implementing the method. Thus, it is advantageous if the calculation of the object shift in the image is carried out to fractions of pixel positions (subpixel-) exactly and the resulting images are optimized by super-resolution methods known to the person skilled in the art in image noise and sharpness. To obtain accuracy is to enlarge all output images by interpolation - about a linear factor 2 or 4, apply to the enlarged images the known pixel-precise algorithms and to reduce the temporally interpolated result images again by the same factor.

Another measure to improve the resulting from interpolation image quality over known methods for video conferencing systems is achieved in that instead of conventional correlation method for calculating the disparities and subsequent filtering and smoothing slower, but more precise global optimizing method for calculating much denser low-noise Disparity values are used. An example of such a useful method is e.g. published in Heiko Hirschmuller, Frank Scholten, Gerd Hirzinger: Stereo Vision Based Reconstruction of Huge Urban Areas from Airborne Pushbroom Camera (HRSC). DAGM Symposium 2005: 58-66.

[62] Another measure to improve the image quality resulting from the interpolation compared to known Driving for videoconferencing systems is achieved by using three or more recordings to calculate the disparity instead of taking two stereo images into account. For three recording positions, the corresponding methods are known as trinocular stereo.

[63] The same applies to the selection of the vertical camera positions of the intermediate images to be reconstructed by interpolation as for the previously described non-equidistant positions of the real sub-cameras, including the formula given above for the vertical position.

[64] d _n = b (cos (2pi * (n + 0, 5) / K)) / 2

[65] d _n in this case denotes the reconstructed vertical position of the n-th image of the periodic image sequence, K the length of the period. This is calculated as indicated by interpolation of the real images from the above and below sub-cameras.

For an unobtrusive 3D rendering, the amplitude b is chosen so that the resulting motion in the image does not exceed 1% of the image height. The number of real sub-cameras in this procedure is at least two. The distance of the outer sub-cameras is chosen so that the vertical camera position of the interpolated sinusoidal vertical movement is always above or at the position of the lowest and lower or on the position of the uppermost sub-camera. Using more than two sub-cameras is advantageous because it reduces the interpolated shifts for the intermediate images and results in less ambiguity and artifacts in the resulting images, especially in complicated occlusion situations where parts of the background are visible only through small recesses in foreground objects or if there are reflections on curved surfaces in the image. In general, the correct representation of the relative displacement of a foreground with recesses of height h requires a stationary approach. background a maximum distance of h / 2 between the vertical positions of the images.

 A particularly flexible design is obtained when geometrically very small cameras, such as wafer-level cameras as sub-cameras in the shortest possible distance in a length of several centimeters are arranged one above the other. By selecting the respectively appropriate for a sequence sub-cameras and possibly additional interpolation in the manner described can thus create a scene adapted vertical oscillating image sequence. On closer inspection, however, shows that the full assembly is not required:

 [67] a particularly advantageous embodiment results when the spacings of the sub-cameras grow progressively from bottom to top (or vice versa), so that e.g. at the positions 0cm, l, 5cmm 3cm, 9cm, 16cm. Subcameras are mounted. With only five sub-cameras, a range of the base width of 0cm to 16cm can be subsequently infinitely realized and the 3D effect can be adapted to the recorded scene. A variable distance adjustment between the sub-cameras for the photographer, as described above, is thereby unnecessary.

[68] It is obvious to use common color sensors in the 3D cameras described here, but the use of color filters is associated with a significant loss of sensitivity compared to pure black / white sensors. Especially with the use of extremely miniaturized wafer-level cameras, this loss leads to noticeable image noise and loss of dynamics. It is therefore advantageous, in partial cameras with miniature optics, not to provide the color filters as a mosaic on the individual sensors, but to multiply the number of wafer-level cameras and to specialize, for example, by using intent filters for specific colors or dynamic ranges. Figure 5 shows an arrangement in which groups of four such single cameras with different color sensitivities (eg entire spectrum (106), red (105), blue (108), green (107) are arranged directly next to each other and four such groups are vertically stacked on top of each other.) The five individual camera groups each generate one color image, the four Groups thus color images of five vertically superimposed recording positions, which are treated in the processing as previously described for recording complete sub-cameras with a single color-suitable optics.

 [69] The extreme miniaturization of the recording elements as wafer-level cameras even allows even greater flexibility for the photographer in that the entire arrangement of five superimposed sub-cameras, each with four individual cameras, with one e.g. 2 times the focal length increased by a factor of 2. The photographer can still choose the focal length he wants to use after shooting. Of course, other forms of processing the single-camera signals are possible, for example, instead of the previously described combination of four cameras each into a color-capable partial camera, e.g. For example, the image signals of the individual cameras, each with the same spectral sensitivity, can also be combined as described above, and the color image can be combined from the results of this calculation step.

[70] The 3D recording and reproducing method according to the invention causes a spatial impression in the observer only because the human perception is able to differentiate even the smallest vertical movements of the displayed image sequence and to interpret them as an object distance. In particular, with a choice according to the invention of a particularly small frequency of the vertical movement of less than 2 Hz for optimum image effect, an undesirable negative effect of this extreme perception sensitivity becomes particularly clear: even more pronounced than in the case of classical stereo reproduction Even the smallest differences in the optical properties of the optics and sensors used in the recording process lead to irritating movement, flickering and wobbly effects.

In order to avoid such irritations, the images of the sub-cameras must be very similar except for the desired vertical displacement or be adjusted by pre-processing each other. The required similarity in brightness and color reproduction and recording time is achieved most advantageously by the use of identical sensors, optics, aperture, focal length and focus and the simultaneous triggering of all sub-cameras with identical exposure time.

[72] Unwanted image motion inaccuracies, which are usually due to production-related unavoidable slightly different lens focal lengths and distortions, can be resolved by compensating for the distortions through geometric preprocessing of the images. Both for the determination of the geometric distortion parameters of the sub-cameras and for the subsequent equalization of the image recordings made with the 3D camera using precisely these previously determined distortion parameters, the person skilled in the art can e.g. Common software programs for so-called "stitching" of panoramic recordings use, for example, by Helmut Dersch, FH Furtwangen, published on the Internet and widely used panoramic tools in conjunction with a front-end such as PTGUI the "New House Internet Services BV".

[73] In contrast to the intended use of such programs, the joining of recorded in different spatial directions single shots here the geometric and, if necessary, color and brightness correction of individual shots in identical recording direction accomplished. However, the determination of the distortion parameters is still done in a known manner via recordings - preferably from distant objects in different spatial directions. The person skilled in the art will use the 3D camera to take pictures of a scene that is full of high-contrast images at a distance of more than 50 m in different directions, for example from the observation tower. In this case, individual images ideally vary by +/- 25% of the image width and height compared to a central recording direction. Finally, he will read in all the images of all sub-cameras in the panorama software and have corresponding pixels automatically determined and possibly nachkorrigieren manually. Subsequently, the expert will gradually release the distortion parameters of each individual sub-camera for the bundle-compensation calculation realized in the software and can thus be calculated. Another possibility for calculating and applying the distortion parameters can be found in the freely available, widely used and open image processing library "Opencv", available at www.willowgarage.com The methods described there are less complicated than the previously described method of using Panorama - Software, because it does not take recordings over distances of 50m, but instead acquires a prefabricated checkerboard pattern over smaller distances, usually less than 5m.

 [74] According to the invention, certain distortion parameters of all sub-cameras are stored in a data memory with this or another method. The distortion parameters consist of the so-called "internal camera parameters", which describe the optical and geometric properties of each individual sub-camera and the "external camera parameters", which represent the position of the cameras in the room - ideally based on the position of one of the sub-cameras.

[75] The contents of the data memory are retrieved later during the processing of recorded images into the 3D sequence and are included in their equalization and color matching in the same way, as in the above-mentioned commercial Pano ramasoftware during the generation of the corrected and corrected partial images.

 [76] Because of the sensitivity of human perception to changes in the image, it is particularly important, as described, that the images of the sub-cameras are taken at exactly the same time. Any movement of objects in the image between shots would otherwise appear as periodic irritations - e.g. Cloud movements between the individual images irritate as periodic forward and backward movements. However, despite this limitation, it may be helpful to include shots that were taken at slightly different times in the calculation of the periodic 3D animation to increase the amount of information in areas of the scene that did not move.

 [77] A possible compromise between the two requirements is that the camera - e.g. consisting of four sub-cameras at a distance of 2 cm each - to mount on a tripod, to take a first picture and then to vertically tilt the camera by e.g. learn to act at height. When calculating the interpolated images, the first 4 images are used as the "master" to compute the vertical motion, and in the areas of the scene where the software detects no change in the first versus the second, the second four shots of the subcameras become additional The use of such a recording method is particularly easy for the photographer, if he does not have to adjust the shift between shots on the tripod, but a vertical adjustment by one or more fixed or adjustable amounts (in the example 1cm) already in the tripod mount the camera or in an adapter between tripod and camera is integrated.

[78] As described above, the method comprises processing spatial 3D sequences, - reading in single images taken from vertically superimposed positions,

 - Equalization and possibly color matching of the individual images, the generation of a sequence of images from all or a subset of the individual images possibly with the addition of interpolated intermediate images

 - Storage on a data medium as an animated sequence of images or as a periodic film.

[79] It is possible to run the steps of the method described successively on different devices for image processing, such as make the recording by a 3D camera and then transfer the image data and calibration parameters to a second device for further processing.

 [80] However, it is particularly advantageous for the method to run in its entirety on the 3D camera, so that the photographer then only has to transfer the periodic image sequence as end product out of the camera. He also has the ability to judge on the monitor of the camera or by a built-in this projector already the effect of the fully edited sequence.

[81] A further advantageous method of implementing the third variant of the method described is to perform the described interpolation of intermediate images only on the playback device. In this case, the playback device receives only the geometrically and optionally color and intensity corrected images of the subcameras and possibly the already calculated disparities and carries out the calculation of the intermediate images independently. The advantage of such a solution is that essential parameters of the reproduction such as frequency and amplitude of the vertical movement need only be set at the time and place of the playback and thus can be adapted to the local conditions. [82] Thus, for example, the frequency and amplitude of the simulated motion during playback on a small image section of the internet presence of a daily newspaper can be selected more aggressively than when presenting a recording on a large monitor in a photography exhibition. Also, it is easier to calculate the same motion pattern on multiple monitors in the same room when calculating the intermediate images on the ground, so as to total one

To create a more soothing atmosphere than when all the monitors reproduced the simulated vertical movements uncoordinated. The real-time calculation of interpolated intermediate images from stereo recordings with a horizontal stereo base is i.a. described in Francesco Isgrö, Emanuele Trucco, Li-Qun Xu, "Towards Teleconferencing by View Synthesis and Large-Baseline Stereo," ICIAP, p. 0198, (ICIAP'01), 2001. More recent work is concerned with the implementation of such techniques mobile terminals, so that one can fall back on the state of the art in the implementation of appropriate playback devices in the selection of the interpolation method.

 [83] The same advantage of a frequency and amplitude of the vertical movement which can be set at the time of reproduction can be achieved by precomputing interpolated intermediate images for a large number of vertically finely graduated positions, if necessary, in another device and storing them in a data memory. Thus, considerably more images are kept available than is necessary for the representation according to the invention with a period of 0.5. 2 Hz. From these pre-maintained intermediate images, the presentation device then retrieves from the memory a sequence of intermediate images which matches the respectively set parameters of frequency and amplitude.

[84] The present document describes by means of several examples the generation of a soft sinusoidal movement by appropriate arrangement of sub-cameras or application of interpolation methods. However, the advantageous spatial perception effect is not only in the representation of a sinusoidal movement given. Rather, other similar forms of exercise may be used. Another equally effective example is given when the movement of the simulated eye movement, for example, is substantially linear and is smoothly transmitted at the reversal points by a parabola.

 [85] The present document describes the advantageous spatial representation on the basis of two or more simultaneous camera shots of a scene. The advantageous representation by a simulated vertical camera movement is analogously applicable to any other implicitly or explicitly present 3D information. Thus, e.g. the scene to be visualized is in the form of a 3D animation or CAD model, which is rendered with a corresponding vertical camera movement.

 [86] A conventional stereo photography or a stereo film can also be used as the starting material if the disparity or distance information is first calculated conventionally from this material, and then this distance information is simulated as in a CAD model for generating the vertical periodic simulations Camera movement is used.

 [87] The method described is suitable not only for the visualization of static objects or photographs but also - as originally intended by the initially quoted VISIDEP method, also for better 3D perception of video or film material. For the creation of the video or. Film materials are different methods applicable:

 [88] An image analogous to the VISIDEP solution with two or more superimposed video cameras. According to the invention, a soft, slow vertical movement between the camera positions is interpolated in the preparation of the material.

[89] For 3D animated films, editing is the easiest way. Here, the virtual Ka meraposition when rendering the film superimposed on a soft vertical movement according to the invention.

 [90] Even conventionally or animated 3D movies can be edited accordingly. In this case, the camera distance of individual picture elements in the scene can be calculated from the stereo information already contained in the material and then a vertical camera movement can be simulated by means of this distance information.

 [91] For some narrow sections of the image information, which are "uncovered" by the simulated vertical camera position, this procedure lacks the necessary information.As with conventional methods of 3D film generation from 2D films, such sites can be identified by so-called "impainting". be filled.

 [92] It is advantageous in the reproduction of spatial video or film material obtained in this way to use the superimposition of a soft vertical movement of the camera position according to the invention only if it is necessary for the generation of a spatial effect. To generate e.g. vertical or lateral camera movements of the underlying material already a generally perfectly sufficient spatial impression of the scene, so that the effect of the invention is usefully used only if the scene has a 3D-Staffeiung and the camera is not

moved in translation.

 [93] The method described here for calculating the moving camera sequence from individual recordings can be implemented as a computer program and stored on a data carrier such as a hard disk, CD or memory card.

Claims

claims 1. A camera for producing photographic image data of a scene, which are suitable for processing with the method according to claims 36 to 53, comprising an optical system, by means of which the scene detected by this detected from different vertically superimposed positions and on at least one image sensor comprising a triggering device, which causes the synchronization of the recording of the image data of the scene imaged by the optics, characterized in that the optics is designed such that the actuation of the triggering device causes a synchronous image recording from three or more vertically superimposed positions. 2. Camera according to claim 1, characterized in that the camera comprises a data memory, which  Contains calibration parameters or distortion parameters, consisting of a subset of internal and external calibration parameters of optics and sensor. 3. A camera for generating photographic image data of a scene, which are suitable for processing with the method according to claims 36 to 53, comprising an optical system, by means of which the scene captured by said scene consists of two different vertical overrides. detected positions and is mapped to at least one image sensor, comprising a triggering device, which causes the synchronization of the recording of the image data of the scene imaged by the optics, characterized in that the camera comprises a data memory containing calibration parameters or distortion parameters, consisting of a subset of internal and external calibration parameters of optics and sensor. 4. A camera according to any one of claims 1 to 3, characterized in that the detection and imaging of the scene from different vertical positions is performed by a single lens, which in particular after the  plenoptic principle is designed, wherein the lens is provided with an optical system by which the light rays incident on the lens distributed on at least n individual lines, where n corresponds to the number of images to be imaged by the lens sub-images, which are assigned to different vertical shooting positions. 5. A camera according to claim 4, characterized in that in front of the image sensor or an imaging plane horizontally arranged lenticular or like-looking optical elements, in particular cylindrical Lenticular are arranged, wherein the aperture of the optics in the horizontal direction is limited so that it corresponds approximately to the effective through the lens grid per frame aperture in the vertical direction. 6. Camera according to one of claims 1 to 3, characterized in that the optics for detecting and imaging the scene from different vertical positions, at least two vertically stacked sub-optics comprises. 7. A camera according to claim 6, characterized in that the vertically stacked sub-optics are formed by sub-cameras (101), which consist of a combination of optics and image sensor. 8. A camera according to claim 7, characterized in that the sub-cameras (101) are not arranged equidistant from each other. 9. A camera according to claim 7 or 8, characterized in that the sub-cameras (101) are formed as independent sub-camera modules at least each consisting of optics and image sensor, wherein the sub-cameras (101) are mechanically connected to each other and by electronic communication means. 10. A camera according to claim 9, characterized in that the sub-camera modules are designed as modules, which can be combined without the use of tools in number and / or geometric arrangement. 11. A camera according to any one of claims 9 or 10, characterized in that the optics of the individual partial camera modules are placed on the narrow side. 12. A camera according to any one of claims 9 to 11, characterized in that the sub-camera modules to the combination of mechanically compatible intermediate pieces, in particular a Dovetail construction have. 13. A camera according to any one of claims 9 to 12, characterized in that a mechanism is provided, by means of which the vertical positioning of the individual sub-camera modules can be varied to each other, in particular locking points for a discrete variation of the positions of the individual sub-camera modules are provided. 14. A camera according to claim 13, characterized in that the mechanism consists of a housing, in which the individual sub-camera modules can be inserted in different vertical positions. 15. Camera according to one of the preceding claims, characterized in that the camera with a device, in particular a tripod, is connected so that the vertical distance to the device by one or more fixed or adjustable amounts can be varied. 16. A camera according to claim 7 or 8, characterized in that the sub-cameras are designed as wafer-level cameras. 17. Camera according to claim 9 - 16, characterized in that the distances of the sub-cameras grow progressively from bottom to top or vice versa. 18. A camera according to any one of claims 9 - 17, characterized in that the sub-cameras are made multiplied and in particular by the use of intent filters on certain colors (105, 106, 107, 108)  and / or dynamic ranges are specialized. Camera according to one of claims 16 to 18, characterized indicates that several arrangements of partial cameras with different focal lengths are installed. 20. A camera according to claim 7 or 8, characterized in that the sub-cameras are integrated together in a single device or a single device component. 21. A camera according to claim 20, characterized in that the sub-cameras are integrated in a flash unit, an adaptable to a hot shoe of a conventional camera housing, a camera tripod, a mobile device or a mobile phone. 22. A camera according to any one of claims 20 or 21, characterized in that the sub-cameras are mounted in a hinged device, in particular distributed over several of its hinged elements, and in particular on the side of these elements, which do not form the outside of the device in the folded state , 23. A camera according to any one of the preceding claims, characterized in that the optics for detecting and imaging the scene from different vertical positions in the manner of a conventional interchangeable lens for mounting on a camera adapted for the reception of interchangeable lenses, in particular a single-lens reflex camera, is formed In particular, the interchangeable lens in the form of partial lenses, the optics of the individual sub-cameras or as a lens attachment carries the corresponding mirror or lens designs, and in particular these partial lenses are arranged along the longitudinal side of the sensor chip. Camera according to one of the preceding claims, characterized in that the optics are connected to a conventional camera via an interface such as a flash contact and / or radio-based communication, in particular based on Bluetooth or WLA. 25. A camera according to claim 24, characterized in that the interface is designed so that are transmitted from the conventional camera, the trigger pulse and / or other recording parameters to the camera. 26. A camera according to any one of claims 24 or 25, characterized in that the interface is designed so that exposure settings of the sub-cameras can be influenced. 27. A camera according to any one of claims 24 to 26, characterized in that the interface is designed so that the image information of the sub-cameras are transmitted to the image memory of the conventional camera. 28. A camera according to any one of claims 24 to 27, characterized in that the conventional camera, which generates conventional 2D images, is independently operable and that the components for spatial imaging and the conventional camera are calibrated to each other. 29. A camera according to any one of claims 7 to 28, that the closure of all sub-cameras consists of a single mechanical unit, in particular of a unit of a plurality of superimposed aperture diaphragms, which use an object side and a common sensor side individual optics. 30. A camera according to any one of claims 7 to 29, that the plurality of sub-cameras is divided into several sets of subsets of sub-cameras, which can be switched between when recording. 31. Camera according to one of the preceding claims, that the triggering device associated trigger is located at the top of the device and in particular right or left of the optics areas are provided for gripping and guiding the camera. 32. Camera according to one of the preceding claims, characterized in that the camera comprises an image data processing unit, which process the recorded image data to a suitable image reproduction image sequence, wherein the image sequence at least combined by the optics in different positions captured images of the scene with each other Images in the image sequence are combined in such a way that, in the sequence, the vertical detection positions which can be assigned to these image data move periodically up and down, 33. Camera according to claim 32, characterized in that the image processing device is integrated together with the scene-imaging optics and the triggering device in a device. 34. A camera according to claim 32, characterized in that the image processing device in a separable from the optics and the triggering device, in particular in a device for generating a sequence of individual images for spatial image reproduction, is integrated. 35. Camera according to one of claims 32 to 34, that the image processing device comprises a device for generating a sequence of individual images for spatial image reproduction. 36. A method for generating an image sequence for a spatial image representation, in which image data are processed by means of an image processing device to a suitable image reproduction image sequence, wherein the image sequence combined image data to which different vertical detection position can be assigned, characterized in that mappings combined in the image sequence in the sequence, the vertical detection positions attributable to these image data move periodically up and down, whereby this periodic motion describes a smooth course in the region of their reversal of direction, and in particular a sinusoidal or in many parts linear and parabolic directional reversal. describes a shaped course. A method according to claim 36, characterized in that the processed image data originate from a camera for generating a spatial image recording, in which is detected by means of an optics detected by this scene from different vertical positions and mapped to at least one image sensor. 38. The method according to claim 36, characterized in that the processed image data are artificially generated and in particular in the form of a three-dimensional animation or CAD model, the image data representing views of the model from different viewing positions. 39. The method according to any one of claims 37 or 38, characterized in that the image data depict a temporal sequence and are supplied in the form of film or video material of the image processing. 40. The method according to any one of claims 37 to 39, characterized in that by means of the image data image sequences are generated which correspond to a superposition of the temporal image sequence of the film or video material with a simulated periodic vertical movement of the camera point. 41. Method according to claim 40, characterized in that the vertical movement of the camera is not simulated for the entire sequence, in particular not if, due to a translatory camera movement, the objects represented in the image data over time already consist of a plurality of observation positions varying in their orientation being represented. 42. The method according to any one of claims 36 to 40, characterized in that when using images from sub-cameras, which in particular by intent filter on special colors are produced, for which further processing by combining the data of the color-specific single camera groups individual color image are generated. 43. The method according to claim 36, wherein not only images mapped directly by the subcameras are used to generate the image sequence, but further mappings are generated from these images by interpolation, which images are also used to generate the image sequence. 44. The method according to claim 43, characterized in that the image noise and the image sharpness are optimized in the context of the interpolation. 45. The method according to any one of claims 43 or 44, characterized in that are used to calculate the disparities between two images three or more mappings. 46. The method according to any one of claims 43 to 45, characterized in that missing or incorrect image data in the generation of individual images are replaced by so-called impainting. 47. Method according to claim 43, characterized in that, in order to eliminate artifacts in subareas of the image data generated by the interpolation for interpolation in these subareas, further image data are used to assign other vertical positions to the original image data used are. 48. The method according to any one of claims 36 to 47, characterized in that the image sequence is designed such that when reproduced the frequency of the periodic course of the individual images to be assigned positions below 2 Hz, preferably at 0.5 Hz. 49. The method according to any one of claims 36 to 48, characterized in that the refresh rate is selected much higher during playback than corresponds to the presentation duration individual fields of the image sequence. 50. The method according to any one of claims 36 to 49, that in the generation of the image sequence images are selected in which results from their most distant assignable positions, a base width at which the resulting vertical movements of pixels in about 1 percent of the image height. 51. The method according to claim 36, characterized in that the images from which the image processing device generates the image sequence are adapted to one another by preprocessing, in particular with respect to their brightness and / or their color rendering and / or to avoid irritating processing. movement, flickering and wobbly effects. 52. The method as claimed in claim 51, characterized in that for the preprocessing of the image data, data about calibration parameters or distortion parameters of the optics, in particular of the subcameras, are read from a data memory and, in particular, the image data are processed so that the distortions of the optics are adjusted for. 53. The method according to any one of claims 36 to 41, characterized indicates that the method is executed on a device other than that used to generate the image data to be processed. 54. An apparatus for generating a spatial reproduction of an image sequence, comprising an interface for receiving image data, and comprising at least one screen or projector for displaying the image sequence, characterized in that the device comprises an image processing device, the received image data, according to the method of one of the claims 36 to 53 processed. 55. Apparatus according to claim 54, characterized  in that the device for receiving the calibration data comprises image data representing the image sequence. 56. Device according to one of claims 54 or 55, characterized in that the device can be connected in common with at least one further device for image reproduction and comprises a synchronizing means for synchronizing the simultaneous reproduction of the image sequence on the connected devices. 57. Device according to one of claims 54 to 56, characterized in that the image processing device is in communication with an image memory, retrieved from the targeted from a superset of individual images and combined to the image sequence to be displayed. 58. Device according to one of claims 54 to 57, characterized in that the device comprises an adjusting means by means of which the amplitude and the frequency of the simulated vertical movement, which results from the sequence of the individual images of the image sequence assignable vertical positions set can be. 59. Computer program product for carrying out a method according to one of claims 36 to 53. Digital storage medium, in particular floppy disk, CD-ROM or DVD, on which a computer program product according to claim 59 is stored.