Television – Panoramic
Reexamination Certificate
2000-03-17
2004-09-14
Diep, Nhon (Department: 2713)
Television
Panoramic
Reexamination Certificate
active
06791598
ABSTRACT:
FIELD OF THE INVENTION
The present invention is related to methods and apparatus for capturing and displaying images and, more particularly, to methods and apparatus for capturing and displaying stereoscopic representations of panoramic images.
BACKGROUND OF THE INVENTION
Ordinary rectilinear photographs provide faithful reproductions of real-life three-dimensional (
3
D) scenes. The resulting rectilinear images, however, appear very flat because much of the depth information, that is, the distances from the camera to the elements of the scene, is lost in the process of capturing and printing these images. Consequently, one simply “looks at” ordinary photographs and there is little sense of being immersed in the scene represented in the photographic image.
Two well-known techniques can be used to enhance the sense of immersion in a photographic or computer generated scene. One of these techniques is stereoscopic imaging in which differing rectilinear images are presented to a viewer's left and right eyes. The differences between these images can be detected by the human visual senses to provide a perception of the relative distances to various elements within a scene. The other technique is panoramic imaging in which a complete 360 degree representation of a scene is captured. Computer software can be used to project an interactively selected portion of such a panoramic image into a rectilinear image, thereby enabling a person to “look around” the scene. The ability to select any view direction interactively over a full 360 degree range provides an alternative sense of immersion. Since stereoscopic images and panoramic images both provide means of allowing a person to experience a sense of immersion in a scene, it would be useful to enhance this sensation further by combining both techniques to allow a person to see stereoscopic views of a panoramic scene.
Numerous devices have been developed for stereoscopic viewing of rectilinear images, including, for example, U.S. Pat. No. 3,850,505 entitled “Universal Stereoscopic Viewing Device,” and U.S. Pat. No. 4,221,462 entitled “Stereoscopic Viewer for Panoramic Camera Film,” the disclosures of which are incorporated by reference herein. Typically, such devices are based on capturing two rectilinear images of a scene, one of which is presented to a viewer's right eye, and the second of which is presented to the viewer's left eye. It should be noted that this process is divided into two distinct steps. In the first step, the scene is captured. In the second step, the captured scene is presented to the viewer. These two steps may take place simultaneously, as with the use of video cameras and video display devices. Alternatively, a persistent medium such as video tape, photographic film, or digital data may be used to capture the scene, allowing the scene to be viewed at a later time.
The scene capture step, as represented in
FIG. 1
, typically involves two cameras separated by some distance (D) perpendicular to the view direction (V). The distance D is defined by the difference between the location of the nodal point
10
of the lens on the right camera (R) and the location of the nodal point
12
of the left camera (L). The point M in
FIG. 1
is the midpoint of the line joining these nodal points. The scene view direction (V) is determined by a vector perpendicular to the line joining the nodal points. The camera view directions (Vleft and Vright) of the left and right cameras are defined by vectors which are perpendicular to the film plane of each camera and which pass through the nodal point of the lens of the corresponding camera. The camera view directions may be parallel to the scene view direction, as shown in
FIG. 1
, or they may be converged to meet and cross in the area of the scene.
In an alternative means of capturing stereoscopic images, a single camera is used. In this case, the camera is placed in the first (left or right) position and a first image is captured. The camera is then moved to the second position where it is used to capture the second image. Otherwise, this method is equivalent to the use of two cameras.
It is also possible to use rectilinear images to capture a panoramic representation of a scene. Such a technique is illustrated in FIG.
2
. This image capture technique may be accomplished by using a single camera to capture a first image (I
1
) with view direction V
1
and field of view, &thgr;
fov
. The field of view angle is determined by the focal length of the lens and the width of the image captured by the camera. The camera may then be rotated by an angle &thgr;
rot
about an axis passing through the nodal point of the lens to obtain a new view direction V
2
, and a second image (I
2
) may be captured with the camera in the new view direction V
2
.
If the rotation angle, &thgr;
rot
, is less than the field of view angle, &thgr;
fov
, then a portion of the left side of image I
2
will overlap a portion of the right side of image I
1
. If image I
2
results from rotating the camera about the nodal point of the lens, then the overlapping portions of images I
1
and I
2
will contain redundant information. In this case, these two images may be projected onto a common surface, such as a cylinder centered on the axis of rotation. The resulting pair of projected images may be merged into a seamless composite image (I
12
) representing a total field of view of (&thgr;
fov
+&thgr;
rot
). This process may be repeated until the resulting field of view encompasses 360 degrees, forming a complete and seamless panoramic image of the scene.
If the camera is rotated about any point other than the nodal point of the lens, then the overlapping portions of images I
1
and I
2
will not contain equivalent representations of the scene. For example, occlusion relations will change, making it possible to see elements of the scene in image I
2
that are not visible in image I
1
. If these images are projected onto a common surface, the overlapping portions of the resulting projected images I
1
′ and I
2
′ will not match and it will not be possible to combine these images into a seamless composite. In this case, it is not possible to produce a seamless panoramic image of the scene.
Consequently, it can be seen that, in order to combine two or more rectilinear images into a seamless panoramic image, the camera must be rotated about the nodal point of the lens, and the camera must not be rotated about an axis which does not pass through the nodal point of the lens.
It is also possible to capture a seamless panoramic image by using a rotating panoramic camera such as that described in U.S. Pat. No. 4,241,985, the disclosure of which is incorporated by reference herein. As with the capture of rectilinear images, this device depends on the ability to rotate about the nodal point of a lens, and the resulting panoramic image corresponds to a projection of the scene onto the surface of a cylinder having a central axis passing through the nodal point of a lens.
Returning to the capture of stereoscopic images, it is possible to attempt to capture a stereoscopic panorama by rotating a pair of cameras about the midpoint (M) between the nodal points, as shown in FIG.
3
. As illustrated, the pair of cameras, L and R, capture images at first positions A, are rotated about the midpoint M, and then capture images at second positions B. In this case, the point M does not coincide with the nodal point of either lens. Consequently, the resulting rectilinear images cannot be combined to form a seamless panoramic image. An example of this procedure has been attempted by Charles Wiltgen (http://www.quicktimefaq.org/). The results of the effort are presented at http://www.quicktimefaq.org/qtvr/archive/Goodies/3Doffice/StereoPano.html, where one can easily see that large image alignment errors are present at the seams between the constituent images.
As an alternative to rotating about the midpoint M, the vector joining the nodal points can be rotated about some other point, such as the nodal point of the left c
Luken William Louis
Shae Zon-Yin
Cameron Douglas W.
Diep Nhon
International Business Machines - Corporation
Ryan & Mason & Lewis, LLP
LandOfFree
Methods and apparatus for information capture and... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Methods and apparatus for information capture and..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Methods and apparatus for information capture and... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3222220