US 11,870,997 B2
Affine motion model derivation method
Jiancong Luo, Skillman, NJ (US); Yuwen He, San Diego, CA (US); and Wei Chen, San Diego, CA (US)
Assigned to VID SCALE, INC., Wilmington, DE (US)
Appl. No. 17/434,974
Filed by VID SCALE, INC., Wilmington, DE (US)
PCT Filed Feb. 28, 2020, PCT No. PCT/US2020/020441
§ 371(c)(1), (2) Date Aug. 30, 2021,
PCT Pub. No. WO2020/180704, PCT Pub. Date Sep. 10, 2020.
Claims priority of provisional application 62/814,125, filed on Mar. 5, 2019.
Prior Publication US 2022/0150506 A1, May 12, 2022
Int. Cl. H04N 19/137 (2014.01); H04N 19/105 (2014.01); H04N 19/149 (2014.01); H04N 19/176 (2014.01)
CPC H04N 19/137 (2014.11) [H04N 19/105 (2014.11); H04N 19/149 (2014.11); H04N 19/176 (2014.11)] 20 Claims
OG exemplary drawing
 
1. A video decoding method comprising:
for at least one current block in a video, determining at least one motion vector gradient from respective motion vectors of a plurality of neighboring sub-blocks neighboring the current block;
determining an estimate of at least one affine parameter for the current block based on an average of q-direction gradients of p-components of motion vectors of the plurality of neighboring sub-blocks, where p and q are each directions selected from the group consisting of x and y;
determining an affine motion model based at least in part on the determined estimate of the affine parameter; and
generating a prediction of the current block using the affine motion model.