Picture-Header Flag for Lower-Cost Bi-Directional Decoding

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The existing solutions in VVC incur unnecessary bit costs due to signaling slice types and bi-directional prediction tools when they are not needed, and it is not straightforward to determine if a coded picture is bi-directional inter coded.

Innovation Solution

Decoding a codeword from the bitstream to indicate the presence of bi-directional inter coded slices, allowing derivation of parameter values to decode segments without explicit signaling of slice types and conditionally enabling bi-prediction tools based on this codeword.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of information

If slice types and bi-directional prediction tools are explicitly signaled in the bitstream, then the decoder can accurately determine picture coding modes, but unnecessary bit costs are incurred when these tools are not needed

Engineering Contradiction:
Improvebit costVSAvoiddetermination of picture coding mode
Core Design Contradiction:
Loss of informationVSEase of operation

Solution Approach 1:

The patent introduces a preliminary flag in the picture header structure that indicates in advance whether bi-directional prediction tools are enabled for the entire picture. This preliminary information allows the decoder to avoid decoding slice type information for all slices, thereby reducing bit cost while maintaining accurate determination of picture coding modes. The flag is set before slice-level decoding occurs, enabling early decision-making about tool usage.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent extracts the bi-directional prediction tool enablement information from the slice-level signaling and places it at the picture header level. This separation allows the slice type signaling to be omitted or simplified when the picture-level flag indicates uniform tool usage across all slices, thereby eliminating redundant information transmission while preserving the ability to accurately determine coding modes.

Inventive Principle:
Principle #2Taking out (Extraction)

2Measurement precision

If slice types are signaled for every slice, then the decoding accuracy is improved, but the bitstream complexity and processing overhead increase

Engineering Contradiction:
Improveslice type determination accuracyVSAvoiddecoding process complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The picture header structure includes a preliminary flag that indicates whether bi-directional prediction tools are enabled for the entire picture. This preliminary information allows the decoder to skip slice type signaling and processing when the coding mode is uniform across all slices, thereby reducing decoding complexity while maintaining accurate determination of slice types through the picture-level indication.

Inventive Principle:
Principle #10Preliminary action

3Adaptability or versatility

If bi-directional prediction tools are enabled for all slices, then decoding flexibility is improved, but bit cost increases due to unnecessary signaling

Engineering Contradiction:
Improvedecoding flexibilityVSAvoidbit cost
Core Design Contradiction:
Adaptability or versatilityVSLoss of information

Solution Approach 1:

A flag in the picture header structure is introduced to preliminarily indicate whether bi-directional prediction tools are enabled for the entire picture. When this flag indicates that the tools are enabled, the decoder can proceed with bi-directional decoding for all slices without requiring individual slice type signaling, thereby maintaining decoding flexibility while avoiding unnecessary bit cost. When the flag indicates tools are disabled, the decoder can skip related processing entirely.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentUS20250301177A1Decoding based on bi-directional picture condition
Publication Date: 2025.09.25 TELEFONAKTIEBOLAGET LM ERICSSON (PUBL)
  • US20250301177A1 patent drawing
  • US20250301177A1 patent drawing
  • US20250301177A1 patent drawing

AI summary

A methods and apparatuses for encoding or decoding a segment in a coded picture. The decoding may include obtaining from a bitstream a value of a first codeword that indicates whether the coded picture may contain bi-directional inter coded segments or not. The decoding may include, based on the value of the first codeword, deriving one or more parameter values by either (a) decoding one or more parameters from the bitstream to obtain the one or more parameter values or (b) inferring the one or more parameter values. The decoding may include decoding the segment in the coded picture based on the one or more parameter values. The encoding may include encoding the first codeword to the bitstream and determining whether to encode one or more parameter values in the bitstream based on a determination of whether the coded picture will contain bi-directional inter coded segments or not.