US 12,244,864 B2
Efficient implementation of matrix-based intra-prediction
Jonathan Pfaff, Berlin (DE); Björn Stallenberger, Berlin (DE); Michael Schafer, Berlin (DE); Philipp Merkle, Berlin (DE); Tobias Hinz, Berlin (DE); Philipp Helle, Berlin (DE); Heiko Schwarz, Berlin (DE); Detlev Marpe, Berlin (DE); Thomas Wiegand, Berlin (DE); Benjamin Bross, Berlin (DE); Martin Winken, Berlin (DE); and Mischa Siekmann, Berlin (DE)
Assigned to Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V., Munich (DE)
Filed by Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V., Munich (DE)
Filed on May 12, 2024, as Appl. No. 18/661,654.
Application 18/661,654 is a continuation of application No. 17/763,624, granted, now 12,022,120, previously published as PCT/EP2020/076697, filed on Sep. 24, 2020.
Claims priority of application No. 19199396 (EP), filed on Sep. 24, 2019.
Prior Publication US 2024/0298033 A1, Sep. 5, 2024
This patent is subject to a terminal disclaimer.
Int. Cl. H04N 19/00 (2014.01); H04N 19/105 (2014.01); H04N 19/132 (2014.01); H04N 19/159 (2014.01); H04N 19/176 (2014.01); H04N 19/46 (2014.01); H04N 19/61 (2014.01)
CPC H04N 19/619 (2014.11) [H04N 19/105 (2014.11); H04N 19/132 (2014.11); H04N 19/159 (2014.11); H04N 19/176 (2014.11); H04N 19/46 (2014.11)] 22 Claims
OG exemplary drawing
 
1. An apparatus for decoding a picture, wherein the apparatus comprises one or more processors configured to:
for a first subset of intra-predicted blocks:
decode a mode index from a data stream, wherein the mode index specifies a prediction matrix from a list of matrix-based intra-prediction modes,
compute a matrix-vector product based on an input vector and the prediction matrix, and
associate components of an output vector obtained from the matrix-vector product with sample positions of the first subset of intra-predicted blocks to obtain a first prediction signal;
for a second subset of intra-predicted blocks, predict the second subset of intra-predicted blocks using one of an angular prediction mode, a DC mode, and a planar prediction mode to obtain a second prediction signal;
for an intra-predicted block, decode a transformation flag from the data stream using context adaptive binary arithmetic coding based on a probability model selected according to whether the intra-predicted block is one of the first subset of intra-predicted blocks or one of the second subset of intra-predicted blocks, wherein the transformation flag indicates whether a prediction residual for the intra-predicted block is coded into the data stream using a default transformation or a different transformation;
decode the prediction residual for the intra-predicted block from the data stream;
transform the prediction residual using a reverse transformation to obtain a prediction residual signal; and
correct the first prediction signal or the second prediction signal using the prediction residual signal.