Granted Patent B2

US 12,450,781 · App. 18/402,142 · Granted Oct 21, 2025

Inventors: Julien Ricard (Plouër-sur-Rance, FR); Celine Guede (Cesson-Sevigne, FR); Yannick Olivier (Thorigne Fouillard, FR); Joan Llach Pinsach (London, GB); David Gendron (Chevaigne, FR)

Assignee: InterDigital VC Holdings, Inc.

G06T9/00H04N19/184H04N19/186

Abstract

Claims (26)

1. A method comprising:

reconstructing at least one current 3D sample of a point cloud using at least one geometry image storing depth values of 3D samples of the point cloud;

determining a maximum depth difference between depth values of neighboring 3D samples of the current 3D sample and a depth value of the current 3D sample; and

adding at least one 3D sample to the point cloud at a same 2 D location as a 2 D sample representing a projection of the current 3 D sample of the point cloud onto an image plane, up to the maximum depth difference.

2. The method of claim 1 , wherein the point cloud is obtained by de-projecting at least one geometry image.

3. The method of claim 1 , wherein a projection of the neighboring 3D samples onto the geometry image and a projection of the current 3D sample onto the geometry image belong to a same patch of the geometry image.

4. The method of claim 1 , wherein a depth value of the at least one 3D sample added to the point cloud belongs to a range from the depth value of the current 3D sample of the point cloud and a depth value being equal to a sum of the depth value of the current 3D sample and the maximum depth difference minus 1.

5. The method of claim 1 , wherein when determining the maximum depth difference between depth values of neighboring 3D samples of the current 3D sample and the depth value of the current 3D sample, a difference between the depth value of the current 3D sample of the point cloud and the depth value of a neighboring 3D sample is ignored if the difference does not belong to a first range of values.

6. The method of claim 1 , wherein the maximum depth difference equals to or is higher than a first value and the first value is 1 or 2.

7. The method of claim 1 , comprising obtaining color information associated with at least one added 3D sample by interpolating color information associated with some other 3D samples of the point cloud.

8. The method of claim 1 , comprising assigning a color-coding mode to the at least one added 3D sample according to the maximum depth difference associated with the added 3D sample, the color-coding mode indicating whether color information associated with the at least one added 3D sample is encoded in a bitstream or not; and wherein responsive to a determination that the color coding mode indicates that the color information is not encoded in the bitstream, color information associated with the at least one added 3D sample is derived from color information associated with at least one 3D sample of the point cloud.

9. The method of claim 1 , wherein a number of the at least one 3D sample added to the point cloud at the same 2D location as the 2D sample representing the projection of the current 3D sample equals the maximum depth difference minus 1.

10. The method of claim 1 , wherein responsive to a determination that the maximum depth difference equals 0 or 1, a single 3D sample is added at the same 2D location as the 2D sample representing the projection of the current 3D sample.

11. A device comprising at least one processor configured to:

reconstruct at least one current 3D sample of a point cloud using at least one geometry image storing depth values of 3D samples of the point cloud;

determine a maximum depth difference between depth values of neighboring 3D sample of the current 3D sample and a depth value of the current 3D sample; and

add at least one 3D sample to the point cloud at a same 2D location as a 2D sample representing a projection of the current 3D sample of the point cloud onto an image plane, up to the maximum depth difference.

12. The device of claim 11 , a depth value of the at least one 3D sample added to the point cloud belongs to a range from the depth value of the current 3D sample of the point cloud and a depth value being equal to a sum of the depth value of the current 3D sample and the maximum depth difference minus 1.

13. The device of claim 11 , wherein when determining the maximum depth difference between depth values of neighboring 3D samples of the current 3D sample and the depth value of the current 3D sample, a difference between the depth value of the current 3D sample of the point cloud and the depth value of a neighboring 3D sample is ignored if the difference does not belong to a first range of values.

14. The device of claim 11 , wherein the at least one processor is further configured to assign a color-coding mode to the at least one added 3D sample, the color-coding mode indicating whether color information associated with the at least one added 3D sample is encoded in a bitstream or not; and wherein responsive to a determination that the color coding mode indicates that the color information is not encoded in the bitstream, color information associated with the at least one added 3D sample is derived from color information associated with at least one 3D sample of the point cloud.

15. The device of claim 11 , wherein a projection of the neighboring 3D samples onto the geometry image and a projection of the current 3D sample onto the geometry image belong to a same patch of the geometry image.

16. The device of claim 11 , wherein the maximum depth difference equals to or is higher than a first value and the first value is 1 or 2.

17. The device of claim 11 , wherein the at least one processor is configured to obtain color information associated with at least one added 3D sample by interpolating color information associated with some other 3D samples of the point cloud.

18. The device of claim 11 , wherein the at least one processor is further configured to, responsive to a determination that the maximum depth difference equals 0 or 1, add a single 3D sample at the same 2D location as the 2D sample representing the projection of the current 3D sample.

19. A non-transitory computer-readable medium including instructions for causing one or more processors to perform reconstructing at least one current 3D sample of a point cloud using at least one geometry image storing depth values of 3D samples of the point cloud, determining a maximum depth difference between depth values of neighboring 3D samples of the current 3D sample and a depth value of the current 3D sample, adding at least one 3D sample to the point cloud at a same 2D location as a 2D sample representing a projection of the curent 3D sample of the point cloud onto an image plane, up to the maximum depth difference.

20. The non-transitory computer-readable medium of claim 19 , wherein a depth value of the at least one 3D sample added to the point cloud belongs to a range from the depth value of the current 3D sample of the point cloud and a depth value being equal to a sum of the depth value of the current 3D sample and the maximum depth difference minus 1.

Assignments (3)

Priority Claims (3)

Continuity (2)

References Cited (44)