Light field camera system and method of setting baseline and convergence distance
A light field camera system and method of setting a baseline and a convergence distance of a multiple camera rig employ sample point depths to determine and set values with respect to a three-dimensional (3D) scene. A baseline and a convergence distance of a multiple camera rig is set by determining a plurality of sample point depths representing distances between the multiple camera rig and a plurality of sample points in the 3D scene within a field of view of the multiple camera rig. The convergence distance is set to be an average sample point depth of the plurality of sample point depths. The baseline is then set as a negative inverse of a difference between a minimum sample point depth of the plurality of sample point depths and the convergence distance. The light field camera system includes a multiple camera rig comprising a plurality of cameras.
1. A method of setting a baseline and a convergence distance of a multiple camera rig, the method comprising:
determining a plurality of sample point depths representing distances between the multiple camera rig and a plurality of sample points in a three-dimensional scene within a field of view of the multiple camera rig;
setting the convergence distance to be an average sample point depth of the plurality of sample point depths;
determining a minimum sample point depth of the plurality of sample point depths; and
setting the baseline to be a negative inverse of a difference between the minimum sample point depth and the convergence distance.
2. The method of setting a baseline and a convergence distance of claim 1 , wherein determining the plurality of sample point depths comprises performing ray casting within the three-dimensional scene and recording hit distances for each sample point of the plurality of sample points.
3. The method of setting a baseline and a convergence distance of claim 1 , wherein determining the plurality of sample point depths comprises computing the sample point depths from a disparity map of the three-dimensional scene for each sample point of the plurality of sample points.
4. The method of setting a baseline and a convergence distance of claim 3 , wherein computing the sample point depths further comprises using image disparity between images captured by different cameras of the multiple camera rig and forming a disparity map of the three-dimensional scene from the image disparity.
5. The method of setting a baseline and a convergence distance of claim 1 , wherein determining the plurality of sample point depths comprises using a depth sensor to measure distances between the multiple camera rig and objects within the three-dimensional scene, the objects corresponding to sample points of the plurality of sample points.
6. The method of setting a baseline and a convergence distance of claim 5 , wherein the depth sensor comprises one of a laser distance sensor and a time-of-flight distance sensor.
7. The method of setting a baseline and a convergence distance of claim 1 , wherein the average sample point depth is a weighted average having weights assigned according to a decreasing function of distance from a focal point within the three-dimensional scene.
8. The method of setting a baseline and a convergence distance of claim 7 , wherein the weights of the weighted average are assigned according to a gaussian distribution centered on the focal point.
9. The method of setting a baseline and a convergence distance of claim 1 , wherein the multiple camera rig comprises a plurality of virtual cameras, the three-dimensional scene being a three-dimensional model.
10. The method of setting a baseline and a convergence distance of claim 1 , wherein the multiple camera rig comprises a plurality of physical cameras, the three-dimensional scene representing a physical scene imaged by cameras of the plurality of physical cameras.
11. The method of setting a baseline and a convergence distance of claim 1 , wherein the multiple camera rig comprises a camera that is moved between a plurality of locations to capture images that form the three-dimensional scene.
12. A light field camera system comprising:
a multiple camera rig comprising a plurality of cameras;
a processor; and
a memory configured to store instructions that when executed by the processor implement:
determining a plurality of sample point depths representing distances between the multiple camera rig and a plurality of sample points in a three-dimensional scene within a field of view of the multiple camera rig; and
setting a convergence distance and a baseline of the multiple camera rig,
wherein the convergence distance is an average sample point depth of the plurality of sample point depths and the baseline is a negative inverse of a minimum sample point depth of the plurality of sample point depths minus the convergence distance.
13. The light field camera system of claim 12 , wherein the plurality of sample point depths is determined one or more of from a depth map of an image representing the three-dimensional scene, using a disparity map to compute the sample point depths, performing ray casting within the three-dimensional scene to determine the sample point depths within the three-dimensional scene.
14. The light field camera system of claim 12 , wherein the plurality of sample point depths is determined using a depth sensor to measure distances between the multiple camera rig and objects within the three-dimensional scene, the objects corresponding to sample points of the plurality of sample points.
15. The light field camera system of claim 12 , wherein the average sample point depth of the plurality of sample point depths is a weighted average, weights of the weighted average being assigned according to a decreasing function of distance from a focal point of the three-dimensional scene.
16. The light field camera system of claim 12 , wherein a camera of the plurality of cameras is a virtual camera, the three-dimensional scene being a three-dimensional model.
17. The light field camera system of claim 12 , wherein cameras of the plurality of cameras of the multiple camera rig comprises physical cameras, the three-dimensional scene representing a physical scene imaged by the physical cameras.
18. The light field camera system of claim 12 , further comprising a multiview display, the convergence distance corresponding to a zero disparity plane of the multiview display.
19. A non-transitory, computer-readable storage medium storing executable instructions that, when executed by a processor of a computer system, performs operations of determining a baseline and convergence distance of a multiple camera rig, the operations comprising:
determining a plurality of sample point depths representing distances between the multiple camera rig and a plurality of sample points in a three-dimensional scene within a field of view of the multiple camera rig, the convergence distance being set as an average sample point depth of the plurality of sample point depths; and
determining a minimum sample point depth of the plurality of sample point depths, the baseline being set as a negative inverse of a difference between the minimum sample point depth and the convergence distance.
20. The non-transitory, computer-readable storage medium of claim 19 , wherein determining the plurality of sample point depths comprises one or more of from a depth map of an image representing the three-dimensional scene, using a disparity map to compute the sample point depths, performing ray casting within the three-dimensional scene to determine the sample point depths within the three-dimensional scene, and using a depth sensor to measure distances between the multiple camera rig and objects within the three-dimensional scene, the objects corresponding to sample points of the plurality of sample points.
21. The non-transitory, computer-readable storage medium of claim 19 , wherein the average sample point depth of the plurality of sample point depths is a weighted average, weights of the weighted average being assigned according to a decreasing function of distance from a focal point of the three-dimensional scene.