Projected Interpolation Prediction for Video Coding Efficiency

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current video coding standards, such as H.264/AVC, are limited in prediction choices, accuracy, and the use of discrete multiple references, leading to inefficient compression and high bitrate requirements for high-resolution video applications.

Innovation Solution

The introduction of Projected Interpolation Prediction (PIP) in the Next Generation Video (NGV) codec, which synthesizes co-located frames from past and future frames to improve prediction accuracy at lower bit costs, reducing prediction errors and signaling costs, and incorporates adaptive techniques to exploit interframe differences beyond motion, such as gain, blur, and registration.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If traditional video coding standards (H.264/AVC) are used, then implementation simplicity is maintained, but compression efficiency is insufficient for high-resolution video applications

Engineering Contradiction:
Improvecompression efficiencyVSAvoidprediction mechanism complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent applies preliminary action by generating a virtual reference picture from future frames before the current frame is encoded. This virtual reference picture is created in advance using motion estimation and pixel projection techniques, allowing the encoder to have more prediction options available before making encoding decisions, thereby improving compression efficiency without significantly increasing decoder complexity

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent introduces a virtual reference picture as an intermediary element between past and future frames. This virtual reference picture serves as a mediator that combines information from multiple reference frames to provide improved prediction for the current frame, enabling better compression while maintaining a relatively simple decoding process

Inventive Principle:
Principle #24Intermediary (Mediator)

2Measurement precision

If more prediction partitions and multiple references are used, then prediction accuracy is improved, but signaling cost and bitrate increase

Engineering Contradiction:
Improveprediction accuracyVSAvoidsignaling cost
Core Design Contradiction:
Measurement precisionVSLoss of information

Solution Approach 1:

The patent uses copying by creating a virtual reference picture that replicates and combines information from multiple future reference frames. Instead of signaling multiple separate motion vectors and reference frame selections, the system creates a single virtual reference picture that encapsulates the predictive information, reducing signaling overhead while maintaining high prediction accuracy

Inventive Principle:
Principle #26Copying

Solution Approach 2:

The patent merges information from multiple reference frames into a single virtual reference picture. By combining motion compensation data from multiple future frames through pixel projection techniques, the system creates one consolidated prediction source that provides high accuracy without requiring separate signaling for each reference frame

Inventive Principle:
Principle #5Merging (Combining)

3Productivity

If discrete multiple references are used, then implementation is simpler, but compression efficiency is limited

Engineering Contradiction:
Improvecompression efficiencyVSAvoidprediction flexibility
Core Design Contradiction:
ProductivityVSAdaptability or versatility

Solution Approach 1:

The patent applies dynamics by creating a virtual reference picture that can adaptively combine information from multiple future reference frames based on the specific content and motion characteristics of each block. This dynamic approach allows the prediction mechanism to flexibly utilize available reference information without requiring complex discrete multiple reference signaling, thereby improving compression efficiency while maintaining implementation feasibility

Inventive Principle:
Principle #15Dynamics

Data Source

PatentEP3087744B1Projected interpolation prediction generation for next generation video coding
Publication Date: 2020.12.02 INTEL CORP
  • EP3087744B1 patent drawingFigure 1
  • EP3087744B1 patent drawingFigure 2
  • EP3087744B1 patent drawingFigure 3(a)

AI summary

Techniques related to projected interpolation prediction generation for video coding are described.