Computer graphics processing and selective visual display system – Computer graphics processing – Three-dimension
Reexamination Certificate
2001-02-07
2004-09-28
Zimmerman, Mark (Department: 2671)
Computer graphics processing and selective visual display system
Computer graphics processing
Three-dimension
C345S420000, C345S424000, C382S154000
Reexamination Certificate
active
06798409
ABSTRACT:
FIELD OF THE INVENTION
This invention relates to the field of 3D displays.
DESCRIPTION OF THE PRIOR ART
Autostereoscopic 3D displays are a well-known class of 3D display that can be viewed without needing to wear special glasses. The general principle is to ensure that the image seen on a display screen is dependant on the position of the viewer. At least two images are usually used, showing the desired view of the scene from two positions, corresponding to the positions of the left and right eye of the viewer for a limited range of head positions. The images are usually created by photographing a scene from positions corresponding to the centres of the intended eye positions. More than two images can be used, which allows a greater degree of head movement, whilst showing changes of parallax in the image, leading to increased realism. The images used for such a so-called multiview display are usually acquired by several cameras, each positioned so as to capture a view of the scene appropriate for the eye position at which the image is visible.
As an alternative to using several cameras, images may be generated using computer graphics techniques to render images equivalent to those that would be captured by a two or more real cameras. The process of rendering an image to simulate an image captured by a real camera is known as perspective projection. Images rendered in this way must of course exhibit the usual properties of images obtained from a conventional camera, specifically that objects appear smaller when they are a long way away, and the angle at which the scene is viewed varies slightly across the image, by an amount equal to the angular width and height of the image.
There are a number of known methods for realising an autostereoscopic display. For example, a sheet of lenticular lenses can be placed over a high-resolution display. The image underneath each lenticular lens is formed by taking one column of pixels from each image in turn, so that the display presents the viewer with each image in sequence as the viewer's head moves parallel to the image. This process is illustrated in FIG.
2
. An example of this “multiview” approach is described in “Design and applications of multiview 3D-LCD” by C. van Berkel et al, 1996 EuroDisplay Conference, Birmingham, October 1996, and “Characterisation and optimisation of 3D-LCD module design” by C. van Berkel and J. Clarke, SPIE International Conference on Electronic Imaging, San Jose, Feb. 11-14, 1997. A set of conventional cameras is used to capture the images.
However, autostereoscopic displays based on these principles do not produce a fully accurate 3D representation of the scene; they merely presents the viewer with a number of discrete views.
One problem is that often, the viewer will be sufficiently close to the screen that one of his eyes is presented with parts of one image from microlenses on the left of the screen and parts of another image from microlenses on the right, due to the changing angle between the screen and the direction of view. These combined views will not represent an accurate portrayal of the scene. This is illustrated in FIG.
4
. It is possible to minimise this effect by careful design of the display system to account for the variation of angle for a given viewing distance. An example of such a design is described in UK patent Application 2,272,597A (Sharp), which describes an arrangement of illumination elements and cylindrical lenses which ensure that a given observer sees a single view across the whole of the display. However, this will only work for observers at a certain range of distances.
There is a known method of generating an autostereoscopic display which makes, no assumptions about the position of the viewer, and is capable of reproducing a 3D scene with high accuracy. This method is known as Integral Imaging, an example of which is described by McCormick et al., “Examination of the requirements for autostereoscopic, full parallax, 3D TV”, IBC '94, IEE Conference Publication 397, September 1994, pp.477-482, and WO-A-9534018. In this method, a special kind of camera captures an image which is ‘replayed’ by placing a sheet of microlenses over the image. The camera consists of an arrangement of lenses, which can be thought of as behaving like an array of very small cameras, one positioned at the location of each microlens. When the image is replayed, each microlens appears to have a colour and brightness that is a function of the angle at which it is viewed. The form of the image underneath the microlens array is such that the overall appearance is of a 3D image. The microlens array should ideally be composed of spherical microlenses, although lenticular lenses can also be used, in which case the image only exhibits parallax in the direction normal to the lenticular elements.
It is known that the image underneath the microlens array can be generated by a computer, rather than by a special kind of camera. This allows a 3D image of a computer model to be generated. However, techniques for doing this have involved ray tracing and simulation of the behaviour of the integral image camera, and are therefore complex and processor-intensive. For example, the image underneath each microlens can be computed by rendering an image of the scene from the viewpoint of that microlens, with an appropriate modification to allow objects that are on both sides of the image plane to be visible. For a typical microlens array consisting of, say 400 by 300 microlenses, this would require 120,000 separate rendering processes.
The present invention seeks to provide—in one of its forms—a method of allowing conventional image rendering hardware to be used efficiently to produce an image to place underneath the microlens array, whilst enabling an accurate 3D representation of the scene to be provided.
SUMMARY OF THE INVENTION
Accordingly, the present invention consists, in one aspect, in a method of providing a viewable representation of a 3D model, comprising the steps of identifying a set of different viewing angles and rendering from said model a set of orthographic projections corresponding respectively with said viewing angles.
It is important to note that the present invention uses orthographic projections. An image rendered using orthographic projection is unlike an image captured with a conventional camera; indeed it is impossible to capture such an image with a conventional camera. One fundamental difference is that the size of an object in an orthographic image projection is independent of its distance from the camera; in conventional perspective projection (or in a normal camera image), more distant objects appear smaller (the so-called perspective effect). This difference comes about because the angle at which the scene is viewed in an orthographic image is constant across the whole image. The field-of-view is specified in terms of the width and height of the rendered image (measured in units of distance, e.g. feet/meters), rather than by a horizontal and vertical field of view (measured in angular units e.g. degrees).
We have found that using orthographic images overcomes a major problem found with conventional multiview displays. Specifically, in such displays, as explained earlier, the property that determines which image is visible from a particular viewing position at any point on the screen is the angle between the screen and the direction of view of that point, as shown in FIG.
4
. This will tend to result in several views being seen across the width of the display, which is an undesirable property if each view corresponds to a normal perspective camera image. As already mentioned, some displays have proposed modifications to the basic arrangements in an attempt to take account of the changing viewing angle across the display, for example the use of varying lens offsets across the display, as described for example in GB 2,272,597A.
In direct contrast to such complex prior methods of compensating for problems with conventional perspective images, the present invention uses orthographic im
Stevens Richard Fawconer
Thomas Graham Alexander
LandOfFree
Processing of images for 3D display does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Processing of images for 3D display, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Processing of images for 3D display will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3229239