Granted Patent B2

US 12,355,977 · App. 18/378,114 · Granted Jul 8, 2025

Inventor: Yu-Chiao Yang (Taipei, TW)

Assignee: SHARP KABUSHIKI KAISHA

H04N19/139H04N19/105H04N19/119H04N19/157H04N19/172H04N19/176H04N19/196

Abstract

Claims (68)

1. A method of decoding video data by an electronic device, the method comprising:

receiving the video data;

determining a block unit from a current frame included in the video data;

determining a plurality of neighboring sub-partitions selected from a plurality of neighboring sub-blocks, wherein:

the plurality of neighboring sub-blocks is included in a plurality of neighboring blocks spatially neighboring the block unit and reconstructed prior to reconstructing the block unit, and

each of the plurality of neighboring sub-partitions is located at a neighboring position;

determining, from the current frame, more than one previously affine-coded block reconstructed in an affine-based mode prior to reconstructing the block unit, wherein:

each of the more than one previously affine-coded block includes a plurality of affine-coded sub-blocks,

each of the plurality of affine-coded sub-blocks is located at an affine position, and

the plurality of affine-coded sub-blocks is different from the plurality of neighboring sub-partitions;

receiving a plurality of neighboring motion vectors of the plurality of neighboring sub-partitions and a plurality of affine motion vectors of the plurality of affine-coded sub-blocks;

deriving a motion linear model based on the plurality of neighboring motion vectors and the plurality of neighboring positions of the plurality of neighboring sub-partitions and the plurality of affine motion vectors and the plurality of affine positions of the plurality of affine-coded sub-blocks; and

determining a plurality of derived motion vectors for the block unit based on the motion linear model for reconstructing the block unit.

2. The method according to claim 1 , wherein the motion linear model is derived using a regression-based motion vector field (RMVF) derivation.

3. The method according to claim 1 , wherein a plurality of model parameters is derived by using the plurality of neighboring motion vectors and the plurality of neighboring positions of the plurality of neighboring sub-partitions and the plurality of affine motion vectors and the plurality of affine positions of the plurality of affine-coded sub-blocks to solve the motion linear model in an error function.

4. The method according to claim 3 , wherein the error function is a mean square error (MSE) function.

5. The method according to claim 1 , wherein:

each of the plurality of neighboring sub-blocks in a specific one of the plurality of neighboring blocks is determined as one of the plurality of affine-coded sub-blocks when the specific one of the plurality of neighboring blocks is reconstructed in the affine-based mode, and

one or more of the plurality of neighboring sub-blocks in the specific one of the plurality of neighboring blocks are determined as the plurality of neighboring sub-partitions and the others of the plurality of neighboring sub-blocks in the specific one of the plurality of neighboring blocks are excluded from the plurality of neighboring sub-partitions when the specific one of the plurality of neighboring blocks is reconstructed in a first prediction mode different from the affine-based mode.

6. The method according to claim 5 , wherein:

each of the one or more of the plurality of neighboring sub-blocks is adjacent to the block unit, and

neighboring sub-blocks other than the one or more of the plurality of neighboring sub-blocks are separated from the block unit by the one or more of the plurality of neighboring sub-blocks.

7. The method according to claim 1 , further comprising:

dividing the block unit into a plurality of sub-block units;

determining a plurality of sub-block positions of the plurality of sub-block units;

deriving one of the plurality of derived motion vectors for each of the plurality of sub-block units by applying a corresponding one of the plurality of sub-block positions to the motion linear model; and

reconstructing each of the sub-block units based on a corresponding one of the plurality of derived motion vectors to reconstruct the block unit.

8. The method according to claim 1 , further comprising:

determining at least one of a plurality of control point (CP) positions of the block unit in the affine-based mode;

determining at least one CP motion vector in the affine-based mode by applying the at least one of the plurality of CP positions to the motion linear model to generate at least one of a plurality of affine candidates; and

reconstructing the block unit based on the at least one of the plurality of affine candidates.

9. The method according to claim 1 , wherein:

a sum of a number of the plurality of neighboring sub-partitions and a number of the plurality of affine-coded sub-blocks for deriving the motion linear model is less than or equal to a quantity threshold.

10. An electronic device for decoding video data, the electronic device comprising:

one or more processors; and

one or more non-transitory computer-readable media coupled to the one or more processors and storing one or more computer-executable instructions that, when executed by at least one of the one or more processors, cause the electronic device to:

receive the video data;

determine a block unit from a current frame included in the video data;

determine a plurality of neighboring sub-partitions selected from a plurality of neighboring sub-blocks, wherein:

the plurality of neighboring sub-blocks is included in a plurality of neighboring blocks spatially neighboring the block unit and reconstructed prior to reconstructing the block unit, and

each of the plurality of neighboring sub-partitions is located at a neighboring position;

determine, from the current frame, more than one previously affine-coded block reconstructed in an affine-based mode prior to reconstructing the block unit, wherein:

each of the more than one previously affine-coded block includes a plurality of affine-coded sub-blocks,

each of the plurality of affine-coded sub-blocks is located at an affine position, and

the plurality of affine-coded sub-blocks is different from the plurality of neighboring sub-partitions;

receive a plurality of neighboring motion vectors of the plurality of neighboring sub-partitions and a plurality of affine motion vectors of the plurality of affine-coded sub-blocks;

derive a motion linear model based on the plurality of neighboring motion vectors and the plurality of neighboring positions of the plurality of neighboring sub-partitions and the plurality of affine motion vectors and the plurality of affine positions of the plurality of affine-coded sub-blocks; and

determine a plurality of derived motion vectors for the block unit based on the motion linear model for reconstructing the block unit.

11. The electronic device according to claim 10 , wherein the motion linear model is derived using a regression-based motion vector field (RMVF) derivation.

12. The electronic device according to claim 10 , wherein a plurality of model parameters is derived by using the plurality of neighboring motion vectors and the plurality of neighboring positions of the plurality of neighboring sub-partitions and the plurality of affine motion vectors and the plurality of affine positions of the plurality of affine-coded sub-blocks to solve the motion linear model in an error function.

13. The electronic device according to claim 12 , wherein the error function is a mean square error (MSE) function.

14. The electronic device according to claim 10 , wherein:

each of the plurality of neighboring sub-blocks in a specific one of the plurality of neighboring blocks is determined as one of the plurality of affine-coded sub-blocks when the specific one of the plurality of neighboring blocks is reconstructed in the affine-based mode, and

one or more of the plurality of neighboring sub-blocks in the specific one of the plurality of neighboring blocks are determined as the plurality of neighboring sub-partitions and the others of the plurality of neighboring sub-blocks in the specific one of the plurality of neighboring blocks are excluded from the plurality of neighboring sub-partitions when the specific one of the plurality of neighboring blocks is reconstructed in a first prediction mode different from the affine-based mode.

15. The electronic device according to claim 14 , wherein:

each of the one or more of the plurality of neighboring sub-blocks is adjacent to the block unit, and

neighboring sub-blocks other than the one or more of the plurality of neighboring sub-blocks are separated from the block unit by the one or more of the plurality of neighboring sub-blocks.

16. The electronic device according to claim 10 , wherein the one or more computer-executable instructions, when executed by the at least one of the one or more processors, further cause the electronic device to:

divide the block unit into a plurality of sub-block units;

determine a plurality of sub-block positions of the plurality of sub-block units;

derive one of the plurality of derived motion vectors for each of the plurality of sub-block units by applying a corresponding one of the plurality of sub-block positions to the motion linear model; and

reconstruct each of the sub-block units based on a corresponding one of the plurality of derived motion vectors to reconstruct the block unit.

17. The electronic device according to claim 10 , wherein the one or more computer-executable instructions, when executed by the at least one of the one or more processors, further cause the electronic device to:

determine at least one of a plurality of control point (CP) positions of the block unit in the affine-based mode;

determine at least one CP motion vectors in the affine-based mode by applying the at least one of the plurality of CP positions to the motion linear model to generate at least one of a plurality of affine candidates; and

reconstruct the block unit based on the at least one of the plurality of affine candidates.

18. The electronic device according to claim 10 , wherein:

a sum of a number of the plurality of neighboring sub-partitions and a number of the plurality of affine-coded sub-blocks for deriving the motion linear model is less than or equal to a quantity threshold.

Assignments (2)

Continuity (2)

References Cited (11)