Signaling Long-Term Reference Pictures in Video Coding

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current video coding techniques, such as those in the HEVC standard, face inefficiencies in signaling long-term reference pictures, particularly when dealing with multiple reference pictures having different least significant bits (LSBs) values, leading to increased bit usage and unnecessary signaling of MSB cycle values.

Innovation Solution

The proposed solution involves improved signaling methods for long-term reference pictures in slice headers, including differential coding of MSB cycle values and selective signaling based on the presence of short-term reference pictures, allowing for more efficient representation of POC values and reduced bit usage by calculating MSB cycle values relative to previously encoded pictures.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of substance

If differential coding of MSB cycle values is used for multiple long-term reference pictures, then bit usage is reduced, but the complexity of decoding increases

Engineering Contradiction:
Improvebit usageVSAvoiddecoding complexity
Core Design Contradiction:
Loss of substanceVSDevice complexity

Solution Approach 1:

The patent applies preliminary action by pre-establishing the differential coding scheme and reference picture ordering before the actual decoding process. The decoder is pre-configured to understand the differential relationship between MSB cycle values, allowing efficient bit usage without requiring complex real-time calculations during decoding.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent introduces an intermediary mechanism by using the base MSB value as a reference point for differential coding. This base value acts as a mediator that simplifies the representation of multiple MSB cycle values, reducing the number of bits needed while maintaining decoding simplicity through a standardized reference framework.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Productivity

If selective signaling based on short-term reference picture presence is implemented, then signaling efficiency is improved, but the device complexity increases

Engineering Contradiction:
Improvesignaling efficiencyVSAvoidsignaling mechanism complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent applies dynamics by making the signaling mechanism adaptive based on the presence or absence of short-term reference pictures. The system dynamically adjusts what information is signaled - when short-term reference pictures are present, different signaling is used compared to when they are absent, optimizing efficiency for each scenario while using manageable complexity through conditional logic.

Inventive Principle:
Principle #15Dynamics

3Measurement precision

If MSB cycle values are signaled for all long-term reference pictures, then reference picture accuracy is maintained, but bit usage increases

Engineering Contradiction:
Improvereference picture accuracyVSAvoidbit usage
Core Design Contradiction:
Measurement precisionVSLoss of substance

Solution Approach 1:

The patent extracts only the necessary information by using differential coding to represent only the differences between MSB cycle values rather than signaling complete values for all long-term reference pictures. This extraction approach maintains reference picture accuracy by preserving the relative relationships while removing redundant information, thereby reducing bit usage significantly.

Inventive Principle:
Principle #2Taking out (Extraction)

Data Source

PatentEP2868081B1Signaling long-term reference pictures for video coding
Publication Date: 2019.04.17 QUALCOMM INC
  • EP2868081B1 patent drawingFigure 1
  • EP2868081B1 patent drawingFigure 2
  • EP2868081B1 patent drawingFigure 3

AI summary

A video decoder may be configured to decode a first value representative of a difference between a base most significant bits (MSBs) value of a picture order count (POC) value of a current picture of video data and a first MSBs value of a first POC value of a first long-term reference picture of the video data, decode a second value representative of a difference between a second MSBs value of a second POC value of a second long-term reference picture of the video data and the first MSBs value, wherein the first POC value and the second POC value have different least significant bits values, and decode at least a portion of a current picture of the video data relative to at least one of the first long-term reference picture and the second long-term reference picture.