US 11,736,701 B2
Hash-based encoder decisions for video coding
Bin Li, Beijing (CN); and Ji-Zheng Xu, Beijing (CN)
Assigned to Microsoft Technology Licensing, LLC, Redmond, WA (US)
Filed by Microsoft Technology Licensing, LLC, Redmond, WA (US)
Filed on Mar. 31, 2021, as Appl. No. 17/219,720.
Application 17/219,720 is a division of application No. 15/508,067, granted, now 11,025,923, previously published as PCT/CN2014/087869, filed on Sep. 30, 2014.
Prior Publication US 2021/0218974 A1, Jul. 15, 2021
Int. Cl. H04N 19/176 (2014.01); H04N 19/142 (2014.01); H04N 19/14 (2014.01); H04N 19/503 (2014.01); H04N 19/154 (2014.01)
CPC H04N 19/176 (2014.11) [H04N 19/14 (2014.11); H04N 19/142 (2014.11); H04N 19/154 (2014.11); H04N 19/503 (2014.11)] 20 Claims
OG exemplary drawing
 
1. A method, implemented by a computing device, for encoding video to produce encoded data, including determining a motion vector (“MV”) precision for a unit of the video, the method comprising:
splitting the unit of the video into multiple blocks to be encoded;
encoding the unit of the video, including:
for each block of the multiple blocks to be encoded, classifying the block into one of a plurality of categories, wherein at least one category of the plurality of categories is a homogeneous category that uses a homogeneous classification criteria, and at least one other category of the plurality of categories is a hash-match category that uses a hash-based matching classification criteria;
based on the classifications for the multiple blocks to be encoded, determining the MV precision for the unit of the video; and
performing motion compensation at the determined MV precision for any of the multiple blocks to be encoded using the motion compensation; and
outputting the encoded data in a bitstream.