Omni-directional stereo system

1. An omnidirectional device capable of collecting horizontal disparities in multiple angles comprising:

a series of prisms and/or mirrors arranged in a circular pattern;

the prisms or mirrors provide a 90 degree shift of the imagery collected between the surfaces of the prisms or mirrors, enabling a single camera to perform the image acquisition;

a second mechanism for collecting the omnidirectional stereo imagery makes use of a single mirror or prism coupled with an active ranging sensor or sensors distributed along the horizontal plane;

the ranging sensors are distributed in a hexagonal pattern around the omnidirectional mirror;

these ranging sensors can make use of ladar, lidar, or structured light to collect the range image; and

this range information is then paired with the image transmitted through the prism and re-rendered as a stereo pair using the post of the display apparatus.

2. The device of claim 1 , wherein a convex catadioptric prisms allow a camera to collect the omnidirectional imagery from multiple view ports that then are used to provide the disparities depending on the Euler angles of the pose of the user stereo display.

3. The device of claim 1 , wherein

convex or concave parabolic or hyperbolic mirrors to perform the same function; and

a first arrangement provides horizontal disparity in two directions, forward and backward.

4. The device of claim 1 , further comprising a second arrangement with six catadioptric mirrors organized in a hexagonal pattern around the 360 degree yaw angles.

5. The device of claim 1 , wherein

selecting images from directly opposing sides of the circle, the system can reconstruct a set of images that are then be provided to the user;

a higher number of prisms are provided around all yaw angles at the cost of having more scenes; and

individual cameras each separate prism or mirror to increase the pixel resolution of the final rendering.

6. The device of claim 1 , wherein the prism geometric construction must provide a 90 degree shift of the imagery collected between the surfaces of the prisms.

7. The device of claim 6 , wherein

modified triangular or pentas prism geometry are used; and

the ray entrance side of the prism is modified with a nonlinear concave or convex surface to increase the field of view of the facet and provide a suitable overlap for blending.

8. The device of claim 1 , wherein

a first column of fiber optic bundles is aligned tangentially to the circle, and a second column of fiber optic bundles is aligned in the opposite direction with respect to the first column; and

the fiber optic bundles are organized to be tangential to a sphere.

9. The device of claim 1 , wherein

arranging the CCD imagers along the tangential surface of the cylinder or sphere; and

a curved imager with individual pixel prisms providing the 90 degree shift of the imagery collected between the surfaces of the prisms.

10. The device of claim 1 , wherein

the omnidirectional stereo apparatus is mounted on a fixed plate;

a laser pointer mounted on a pan-tilt head is used to project light to a panel of known location;

the resulting image is collected providing the location of the particular pixel in the image frame;

since the location of the panel is known and the location of the pointer is known, and the location of the projected laser beam is also known, the two known locations are used to compute the unit vector corresponding for each pixel on the image; and

this collection of unit vectors that map the particular image pixel to array and space will be used by the rendering system to figure out the display using simple geometrical transformations.

11. The device of claim 1 , wherein

the sensing apparatus is placed on a pan-tilt head while maintaining the projected dot in a known location;

the sensing apparatus is comprised of two or more back to back camera pairs;

to speed up the calibration process, the number of pixels measured is a sub sample of the complete set;

linear and nonlinear approximators are used to interpolate the subsample set; and

if the model of the individual optical devices (prisms or mirrors) is known, the complete or partial set is used to optimize the model parameters using standard nonlinear optimization methods.

12. The device of claim 11 , wherein

the range sensors are calibrated by swapping the laser dot with a morphological feature that is recovered by the range sensors; and

once such morphological feature is a hole or drop off in the panel that is sensed as a range discontinuity on the ladar imager.

13. The device of claim 1 , wherein a quasi-concentric image spheres method,

the imagery collected from the prisms is geometrically transformed and blended into two image spheres; and

each corresponding pixel on both spheres is selected to provide the physical disparity closely related to the disparity of the physical display apparatus.

14. An omnidirectional device capable of collecting horizontal disparities in multiple angles comprising:

a series of prisms and/or mirrors arranged in a circular pattern;

the prisms or mirrors provide a 90 degree shift of the imagery collected, enabling a single camera to perform the image acquisition;

wherein texture and range streams use a different encoding mechanism that is composed of three different streams,

the first stream provides an omnidirectional image from one of the viewpoints;

a second stream provides a range map for each pixel in the sphere;

a third stream provides a list of occluded fillets both in range and in texture;

these are the surfaces that are occluded on the left image, but become visible by the right eye (or head motion);

providing these hidden surfaces from the left eye with sufficient margin to cover the image disparity created by the eye separation and motion of the head, the right eye image is synthesized;

this reconstruction occurs using the range information, the left eye information, and the fillets;

encoding information this way will be an approximation in the sense that all of the range information will be provided by the morphological disparity encoded into the range maps, but changes in the texture of the surfaces created by the eye separation will not be taken under consideration, this will not affect non reflective surfaces, but it will affect mirrors in the scene.

15. The device of claim 8 , wherein

local rendering of reflecting surfaces is accomplished using the standard video gaming techniques; and

these reflections are added to the left eye textures.

16. The device of claim 1 or 14 , wherein using Direct Encoding by Symmetric Bi-sections,

the collection apparatus have natural image partitions created by the number of prisms or mirrors used;

because the prisms are mounted to a fixed surface, images in each individual prism will rotate and move in unison;

because of the inversions and symmetries created by the prisms, the apparent motion of the image as seen by the single camera will be in different directions, therefore confusing standard encoding techniques;

bisecting and rotating the subsections of the images, the images of each prism are aligned, flipped, and rotated to allow standard mpeg encoding techniques to be effective; and

additional mpeg algorithms are implemented where the known correlation between the motion vectors of different parts of the image are exploited for bisecting and rotating the subsections of the images.

17. The device of claim 1 or 14 , wherein

in a first rendering embodiment, two quasi-concentric image spheres are initially generated;

given the Euler angles of the display apparatus, the spheres are cropped to retrieve the view port required by the display; and

one sphere provides the right image, the second sphere provides the left image.

18. The device of claim 1 or 14 , wherein

the image for the left eye is a direct crop if no head motion is considered, however, the image for the right needs to be generated;

given that the range image is available, each pixel is rotated using a graphic engine;

the rotation will expose areas that where originally hidden; and

these hidden surfaces are recovered from the list of fillets available a third stream providing a list of occluded fillets both in range and in texture.

19. The device of claim 1 or 14 , wherein

the rendering device will need to perform the real-time blending of the images and cropping for both eyes; and

only the portion of the sphere that will be displayed to the user.