Video Decoding Reference Picture List Structure Management

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In the context of video coding and decoding, particularly in "Versatile Video Coding (Draft 8)", there are issues with the reference picture list structure, where a structure with zero reference pictures can lead to an indefinite reference picture, and the number of reference pictures used for prediction is not consistently defined across picture and slice headers, resulting in redundancy in weighted prediction descriptions.

Innovation Solution

A video decoding apparatus is configured to manage multiple reference picture list structures, allowing selection on a per-picture or per-slice basis. This configuration ensures that the number of reference pictures is defined and not indefinite, and it allows for the correct definition of the number of reference pictures in both picture and slice headers, thereby correcting redundancy in weighted prediction information.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If a reference picture list structure with zero reference pictures is defined, then the structure can be simplified, but the reference picture becomes indefinite

Engineering Contradiction:
Improvereference picture list structureVSAvoidreference picture definition
Core Design Contradiction:
Device complexityVSReliability

Solution Approach 1:

The patent introduces a parameter `num_ref_pic` that specifies the number of reference pictures in the reference picture list structure. By changing the structure to include this explicit parameter, the system prevents indefinite reference pictures while maintaining structural simplicity. The parameter ensures that at least one reference picture is always defined, resolving the contradiction between structural simplicity and reference picture reliability.

Inventive Principle:
Principle #35Parameter changes

2Adaptability or versatility

If the number of reference pictures is defined in slice header but not in picture header, then flexibility is improved, but consistency and reliability deteriorate

Engineering Contradiction:
Improvereference picture configuration flexibilityVSAvoidreference picture definition consistency
Core Design Contradiction:
Adaptability or versatilityVSReliability

Solution Approach 1:

The patent segments the reference picture list structure into multiple instances that can be independently configured at different hierarchy levels (picture header and slice header). Each segment can have its own `num_ref_pic` parameter, allowing flexible configuration while maintaining consistency through hierarchical inheritance. The picture header provides default values that are inherited by slice headers, ensuring both flexibility and reliability.

Inventive Principle:
Principle #1Segmentation

3Measurement precision

If the number of weights is explicitly defined in weighted prediction, then prediction accuracy is improved, but the description becomes redundant when the number of reference pictures is already known

Engineering Contradiction:
Improveprediction accuracyVSAvoidweighted prediction description
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent makes the `num_ref_pic` parameter serve multiple functions: it defines the number of reference pictures for prediction and simultaneously determines the number of weights in weighted prediction. This multi-functionality eliminates redundancy by using a single parameter to control both aspects, while still allowing explicit weight definitions when needed for enhanced prediction accuracy.

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

Data Source

PatentUS20250056039A1Video decoding apparatus
Publication Date: 2025.02.13 SHARP KK
  • US20250056039A1 patent drawing
  • US20250056039A1 patent drawing
  • US20250056039A1 patent drawing

AI summary

An apparatus includes an inter prediction unit configured to decode multiple reference picture list structures and to select one reference picture list structure from the multiple reference picture list structures on a per picture basis or on a per slice basis, wherein in the multiple reference picture list structures, the number of all reference pictures is one or more.