Method and apparatus for processing video signals using coefficient induced prediction

Patent metadata

Field	Value
Publication number	US-10771815-B2
Application number	US-201615764599-A
Country	US
Kind code	B2
Filing date	Sep 29, 2016
Priority date	Sep 29, 2015
Publication date	Sep 8, 2020
Grant date	Sep 8, 2020

Abstract

The present invention provides a method for encoding a video signal on the basis of a graph-based lifting transform (GBLT), comprising the steps of: detecting an edge from an intra residual signal; generating a graph on the basis of the detected edge, wherein the graph includes a node and a weight link; acquiring a GBLT coefficient by performing the GBLT for the graph; quantizing the GBLT coefficient; and entropy-encoding the quantized GBLT coefficient, wherein the GBLT includes a partitioning step, a prediction step, and an update step.

First claim

The invention claimed is:

1. A method of encoding a video signal based on a graph-based lifting transform (GBLT), comprising:
detecting an edge from an intra residual signal, wherein a model for the intra residual signal is designed by using a Gaussian Markov Random Field (GMRF);
generating a graph based on the edge, wherein the graph comprises a node and a weight link;
obtaining a GBLT coefficient by performing the GBLT for the graph;
quantizing the GBLT coefficient; and
entropy-encoding the quantized GBLT coefficient,
wherein the GBLT comprises a split process, a prediction process, and an update process,
wherein the split process of the GBLT is performed to minimize a Maximum A Posteriori (MAP) estimate error within a prediction set, and
wherein the method further comprises:
obtaining a DCT coefficient by performing a DCT on the intra residual signal;
comparing a rate-distortion cost of the DCT coefficient with a rate-distortion cost of the GBLT coefficient;
determining, based on the rate-distortion cost of the GBLT coefficient being smaller than the rate-distortion cost of the DCT coefficient, a mode index corresponding to the GBLT; and
entropy-encoding the mode index.

2. The method of claim 1 , wherein the split process comprises:
calculating a size of an update set;
selecting a node minimizing an MAP estimate error within a prediction set based on the size of the update set; and
calculating an update set for the selected node.

3. The method of claim 1 , wherein the graph is reconnected prior to a next GBLT.

4. A method of decoding a video signal based on a graph-based lifting transform (GBLT), comprising:
extracting a mode index indicative of a transform method from the video signal;
deriving a transform corresponding to the mode index, wherein the transform indicates one of a DCT and the GBLT;
performing an inverse transform for an intra residual signal based on the transform; and
generating a reconstructed signal by adding the inverse-transformed intra residual signal to a prediction signal,
wherein a model for the intra residual signal is designed by using a Gaussian Markov Random Field (GMRF),
wherein the mode index is determined by comparing a rate-distortion cost of a DCT coefficient with a rate-distortion cost of a GBLT coefficient, and
wherein a split process of the GBLT is performed to minimize a Maximum A Posteriori (MAP) estimate error within a prediction set.

5. An apparatus for encoding a video signal based on a graph-based lifting transform (GBLT), comprising:
a processor configured to:
detect an edge from an intra residual signal, wherein a model for the intra residual signal is designed by using a Gaussian Markov Random Field (GMRF);
generate a graph based on the detected edge and obtaining a graph-based lifting transform (GBLT) coefficient by performing the GBLT for the graph;
quantize the GBLT coefficient; and
perform entropy encoding for the quantized GBLT coefficient,

wherein the GBLT comprises a split process, a prediction process, and an update process,
wherein the split process of the GBLT is performed to minimize a Maximum A Posteriori (MAP) estimate error within a prediction set, and
wherein the processor is further configured to:
obtain a DCT coefficient by performing a DCT on the intra residual signal;
compare a rate-distortion cost of the DCT coefficient with a rate-distortion cost of the GBLT coefficient;
determine, based on the rate-distortion cost of the GBLT coefficient being smaller than the rate-distortion cost of the DCT coefficient, a mode index corresponding to the GBLT; and
entropy-encode the mode index.

6. An apparatus for decoding a video signal based on a graph-based lifting transform (GBLT), comprising:
a processor configured to:
extract a mode index indicative of a transform method from the video signal;
derive a transform corresponding to the mode index and perform an inverse transform for an intra residual signal based on the transform; and
generate a reconstructed signal by adding the inverse-transformed intra residual signal to a prediction signal,

wherein a model for the intra residual signal is designed by using a Gaussian Markov Random Field (GMRF), and
wherein the transform indicates one of a DCT and the GBLT,
wherein the mode index is determined by comparing a rate-distortion cost of a DCT coefficient with a rate-distortion cost of a GBLT coefficient, and
wherein a split process of the GBLT is performed to minimize a Maximum A Posteriori (MAP) estimate error within a prediction set.

Assignees

• Lg Electronics Inc
• Univ Southern California

Inventors

• Chao Yung-Hsuan
• Yea Sehoon
• Ortega Antonio

Classifications

• H04N19/196

  being specially adapted for the computation of encoding parameters, e.g. by averaging previously computed encoding parameters (processing of motion vectors H04N19/513) · CPC title

• H04N19/107

  between spatial and temporal predictive coding, e.g. picture refresh · CPC title

• H04N19/13

  Adaptive entropy coding, e.g. adaptive variable length coding [AVLC] or context adaptive binary arithmetic coding [CABAC] · CPC title

• H04N19/11

  among a plurality of spatial predictive coding modes · CPC title

• H04N19/96

  Tree coding, e.g. quad-tree coding · CPC title