Video Block Transform Selection With Inferred LFNST Signaling

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The existing video codec standardization process in Versatile Video Coding (VVC) involves extensive testing and signaling of multiple transform selection (MTS) and low frequency non-separable transform (LFNST) combinations, leading to increased encoder complexity and bit usage, without necessarily improving compression efficiency.

Innovation Solution

By limiting the number of MTS and LFNST combinations evaluated and reducing the need for signaling, the encoder and decoder complexity is reduced, with specific transform combinations being inferred or omitted based on predetermined indices, thereby optimizing the transform selection process.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If the encoder tests all allowed combinations of MTS and LFNST transforms, then the compression efficiency is improved, but the encoder complexity and run time increase significantly

Engineering Contradiction:
Improvecompression efficiencyVSAvoidencoder complexity
Core Design Contradiction:
Manufacturing precisionVSDevice complexity

Solution Approach 1:

The patent extracts and removes unlikely transform combinations from the evaluation set. Specifically, it identifies and eliminates combinations where LFNST is applied with certain MTS transforms based on predefined criteria, thereby reducing the number of combinations the encoder must test while preserving the most effective transformations for compression.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

Instead of evaluating all possible transform combinations, the patent applies partial evaluation by limiting the search to a subset of combinations. The encoder performs transform selection only for the most promising combinations identified through predefined rules, rather than exhaustively testing every possible combination of MTS and LFNST transforms.

Inventive Principle:
Principle #16Partial or excessive action

2Reliability

If the encoder signals all possible MTS and LFNST combinations, then the decoder can accurately reconstruct the transform applied, but the bit usage increases

Engineering Contradiction:
Improvetransform reconstruction accuracyVSAvoidbit usage
Core Design Contradiction:
ReliabilityVSQuantity of substance

Solution Approach 1:

The patent removes the need to signal certain transform combinations by inferring them at the decoder side. When specific MTS transforms are detected, the corresponding LFNST application is inferred without explicit signaling, thereby reducing the number of bits required while maintaining accurate transform reconstruction.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The decoder performs self-service by automatically inferring which LFNST transforms were applied based on the received MTS transform information. This eliminates the need for separate signaling bits, as the decoder can independently determine the transform combination used during encoding through predefined inference rules.

Inventive Principle:
Principle #25Self-service

3Manufacturing precision

If the encoder evaluates all MTS and LFNST combinations, then the best compression is achieved, but the memory usage and processing time increase

Engineering Contradiction:
Improvecompression efficiencyVSAvoidencoder run time
Core Design Contradiction:
Manufacturing precisionVSLoss of time

Solution Approach 1:

The patent extracts and removes computationally expensive transform combinations from the evaluation process. By identifying and eliminating combinations that are unlikely to provide optimal compression based on predefined criteria, the encoder reduces the number of transforms that need to be computed and evaluated, thereby reducing run time while maintaining compression quality.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent performs preliminary filtering of transform combinations before the actual encoding process. By pre-defining which combinations are worth evaluating and which can be skipped, the encoder prepares a reduced set of candidates in advance, avoiding unnecessary computational work during the time-critical encoding phase.

Inventive Principle:
Principle #10Preliminary action

4Adaptability or versatility

If multiple LFNST transform sets are maintained for different intra prediction directions, then the adaptability is improved, but the device complexity increases

Engineering Contradiction:
Improvetransform selection flexibilityVSAvoiddecoder complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent makes the decoder universal by enabling it to handle multiple transform combinations through a single unified process. Instead of requiring separate decoding paths for different LFNST sets, the decoder uses a general-purpose inference mechanism that works for all transform types, thereby maintaining adaptability while reducing complexity.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentEP3987801B1Method and apparatus for non-separable transform selection in an encoder and decoder
Publication Date: 2026.01.14 TELEFONAKTIEBOLAGET LM ERICSSON (PUBL)
  • EP3987801B1 patent drawingFigure 1
  • EP3987801B1 patent drawingFigure 2~4
  • EP3987801B1 patent drawingFigure 5

AI summary

A method, decoder apparatus, and computer programs are provided. A bit stream having a video block encoded with a multiple transform selection (MTS) index is received. The MTS index is parsed to determine a value of it. Responsive to the MTS index value being a predetermined value (or one of a predetermined set of indices), a LFNST index is parsed from the bit stream to determine a value of it, the value providing an indication whether a LFNST transform set is to be used in decoding the video block. Responsive to the MTS index value not being the predetermined value (or one of the set of indices), the value of the LFNST index is set to a default value indicating no LFNST transform is to be used in the decoding. The video block is decoded based on the values of the MTS index and the LFNST index.