Upsampling low temporal resolution depth maps
Systems and methods are provided for upsampling low temporal resolution depth maps. This upsampling is performed by obtaining a stereo pair of images of a scene captured at a first timepoint, generating a first depth map of the scene for the first timepoint by performing stereo matching on the stereo pair of images, obtaining a subsequent stereo pair of images captured at a subsequent timepoint to the first timepoint, and generating a subsequent depth map that corresponds to the subsequent timepoint by applying an edge-preserving filter using the first depth map without performing stereo matching on the subsequent stereo pair of images.
1. A system for upsampling low temporal resolution depth maps to provide depth maps at low computational cost, comprising:
one or more processors; and
one or more hardware storage devices having stored computer-executable instructions that are operable, when executed by the one or more processors, to cause the system to:
obtain a stereo pair of images of a scene captured at a first timepoint;
generate a first depth map of the scene for the first timepoint by performing stereo matching on the stereo pair of images;
obtain a subsequent stereo pair of images captured at a subsequent timepoint to the first timepoint; and
generate a subsequent depth map that corresponds to the subsequent timepoint by applying an edge-preserving filter using the first depth map without performing stereo matching on the subsequent stereo pair of images, thereby reducing computational cost associated with generating the subsequent depth map, wherein the edge-preserving filter utilizes (i) the first depth map, (ii) least one image of the stereo pair of images, (iii) at least one image of the subsequent stereo pair of images, and (iv) an additional depth map or at least one image of an additional stereo pair of images to generate the subsequent depth map.
2. The system of claim 1 , further comprising:
a stereo pair of cameras, wherein the stereo pair of cameras captures the stereo pair of images and the subsequent stereo pair of images.
3. The system of claim 1 , wherein the edge-preserving filter is a joint bilateral filter.
4. The system of claim 1 , wherein the additional stereo pair of images and the additional depth map are associated with a timepoint subsequent to the subsequent timepoint.
5. The system of claim 1 , wherein the additional stereo pair of images and the additional depth map are associated with a timepoint prior to the first timepoint.
6. The system of claim 1 , wherein the computer-executable instructions are further operable, when executed by the one or more processors, to cause the system to:
successively obtain a plurality of additional stereo pairs of images at a first frequency; and
successively generate a first plurality of additional depth maps at a second frequency by performing stereo matching on some additional stereo pairs of images of the plurality of additional stereo pairs of images.
7. The system of claim 6 , wherein the computer-executable instructions are further operable, when executed by the one or more processors, to cause the system to:
successively generate a second plurality of additional depth maps at the first frequency by applying an edge-preserving filter using the additional depth maps of the first plurality of additional depth maps.
8. The system of claim 6 , wherein the first frequency is within a range of 45 to 90 Hz and the second frequency is within a range of 15 to 30 Hz.
9. The system of claim 8 , wherein the second frequency is dynamically updated based on a change in a position of the system relative to the scene.
10. The system of claim 9 , wherein the change in the position of the system relative to the scene is based on tracking data that indicate a pose of the system.
11. The system of claim 9 , wherein the change in the position of the system relative to the scene is based on a difference between at least two depth maps of the second plurality of additional depth maps.
12. The system of claim 1 , wherein the computer-executable instructions are further operable, when executed by the one or more processors, to cause the system to:
reproject depth points based on the subsequent depth map to correspond to a user perspective.
13. A method for upsampling low temporal resolution depth maps to provide depth maps at low computational cost, comprising:
obtaining a stereo pair of images of a scene captured at a first timepoint;
generating a first depth map of the scene for the first timepoint by performing stereo matching on the stereo pair of images;
obtaining a subsequent stereo pair of images captured at a subsequent timepoint to the first timepoint; and
generating a subsequent depth map that corresponds to the subsequent timepoint by applying an edge-preserving filter using the first depth map without performing stereo matching on the subsequent stereo pair of images, thereby reducing computational cost associated with generating the subsequent depth map, wherein the edge-preserving filter utilizes (i) the first depth map, (ii) least one image of the stereo pair of images, (iii) at least one image of the subsequent stereo pair of images, and (iv) an additional depth map or at least one image of an additional stereo pair of images to generate the subsequent depth map.
14. The method of claim 13 , wherein the additional stereo pair of images and the additional depth map are associated with a timepoint subsequent to the subsequent timepoint.
15. The method of claim 13 , wherein the additional stereo pair of images and the additional depth map are associated with a timepoint prior to the first timepoint.
16. The method of claim 13 , further comprising:
successively obtaining a plurality of additional stereo pairs of images at a first frequency; and
successively generating a first plurality of additional depth maps at a second frequency by performing stereo matching on some additional stereo pairs of images of the plurality of additional stereo pairs of images.
17. The method of claim 16 , further comprising:
successively generating a second plurality of additional depth maps at the frequency by applying an edge-preserving filter using the additional depth maps of the first plurality of additional depth maps.
18. One or more hardware storage devices having stored thereon computer-executable instructions, the computer-executable instructions being executable by one or more processors of a computer system to cause the computer system to:
obtain a stereo pair of images of a scene captured at a first timepoint;
generate a first depth map of the scene for the first timepoint by performing stereo matching on the stereo pair of images;
obtain a subsequent stereo pair of images captured at a subsequent timepoint to the first timepoint; and
generate a subsequent depth map that corresponds to the subsequent timepoint by applying an edge-preserving filter using the first depth map without performing stereo matching on the subsequent stereo pair of images, thereby reducing computational cost associated with generating the subsequent depth map, wherein the edge-preserving filter utilizes (i) the first depth map, (ii) least one image of the stereo pair of images, (iii) at least one image of the subsequent stereo pair of images, and (iv) an additional depth map or at least one image of an additional stereo pair of images to generate the subsequent depth map.