Reference Picture List Construction for Multi-View Video Coding

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current video coding standards, such as HEVC, face challenges in efficiently constructing and modifying reference picture lists, particularly when handling inter-view reference pictures, which can lead to increased bit costs and inefficient placement of reference pictures in the list.

Innovation Solution

The proposed solution involves modifying the syntax elements in slice headers to allow for the construction and modification of reference picture lists based on layer identifiers, enabling the efficient placement of inter-view reference pictures within the lists, and signaling their position through additional syntax elements in the slice headers.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If current video coding standards are used to construct reference picture lists, then video coding can be performed, but inter-view reference pictures cannot be efficiently placed in the lists leading to increased bit costs

Engineering Contradiction:
Improvevideo coding efficiencyVSAvoidbit cost
Core Design Contradiction:
ProductivityVSLoss of information

Solution Approach 1:

The reference picture list construction process is segmented into distinct phases: initializing the list with temporal reference pictures, and then conditionally adding inter-view reference pictures through separate syntax elements. This segmentation allows independent optimization of each phase, reducing unnecessary bit costs while maintaining coding efficiency.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The reference picture list is pre-initialized with temporal reference pictures before inter-view reference pictures are considered. This preliminary action establishes a baseline structure that reduces the complexity of subsequent modifications and minimizes the bit cost for indicating reference picture positions.

Inventive Principle:
Principle #10Preliminary action

2Adaptability or versatility

If inter-view reference pictures are added to reference picture lists, then 3D video coding capability is enabled, but the complexity of constructing and modifying the lists increases

Engineering Contradiction:
Improve3D video coding capabilityVSAvoidreference picture list construction complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The reference picture list structure is made dynamic through conditional syntax elements that allow the list to adapt its composition based on the presence of inter-view reference pictures. The list can be initialized with temporal references and then dynamically extended with inter-view references only when needed, reducing unnecessary complexity.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

Different parts of the reference picture list are treated with different qualities: temporal reference pictures form the core structure with standardized handling, while inter-view reference pictures are added as optional extensions with specialized syntax elements only where needed, reducing overall system complexity.

Inventive Principle:
Principle #3Local quality

3Manufacturing precision

If syntax elements are added to slice headers for inter-view reference pictures, then reference picture placement is improved, but the overhead of the slice header increases

Engineering Contradiction:
Improvereference picture placement precisionVSAvoidslice header overhead
Core Design Contradiction:
Manufacturing precisionVSLoss of information

Solution Approach 1:

The syntax elements in slice headers use parameter changes to control reference picture placement: conditional flags indicate whether inter-view reference pictures are present, and position indicators specify their locations only when needed. This parameter-based approach achieves precise placement control while minimizing overhead through conditional coding.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP2885917B1Constructing reference picture lists for multi-view or 3dv video coding
Publication Date: 2021.11.03 QUALCOMM INC
  • EP2885917B1 patent drawingFigure 1
  • EP2885917B1 patent drawingFigure 2
  • EP2885917B1 patent drawingFigure 3

AI summary

In one example, a video coder, such as a video encoder or decoder, is configured to code a value for a layer identifier in a slice header for a current slice in a current layer of multi-layer video data, and, when the value for the layer identifier is not equal to zero, code a first set of syntax elements in accordance with a base video coding standard, and code a second set of one or more syntax elements in accordance with an extension to the base video coding standard. The second set of syntax elements may include a syntax element representative of a position for an identifier of an inter-layer reference picture of a reference layer in a reference picture list, and the video coder may construct the reference picture list such that the identifier of the inter-layer reference picture is located in the determined position.