Intra-block copy prediction for video encoding

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing video coding techniques face challenges in reducing memory bandwidth and storage requirements while maintaining high video quality, particularly with the increased demand for high-resolution and high-fidelity video data, as seen in the inefficiencies of intra-block copy prediction methods which often exceed the memory bandwidth limits of HEVC standards.

Innovation Solution

The implementation of selective intra-block copy prediction techniques that disable inter-picture bi-prediction or uni-prediction when certain conditions are met, such as smaller prediction block sizes or specific flag indications, to reduce memory bandwidth and size requirements, and the use of surrogate reference samples to restrict reference block usage.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If intra-block copy prediction is used to improve coding efficiency, then video quality is improved, but memory bandwidth requirements exceed HEVC limits

Engineering Contradiction:
Improvevideo qualityVSAvoidmemory bandwidth
Core Design Contradiction:
Manufacturing precisionVSQuantity of substance

Solution Approach 1:

The patent segments the prediction process by disabling bi-prediction for certain block sizes and introducing a restriction flag. This divides the prediction modes into separate categories (uni-prediction for restricted blocks, bi-prediction for others), allowing memory bandwidth to be controlled while maintaining coding efficiency for blocks that benefit from intra-block copy prediction

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent changes the parameter of prediction mode selection by introducing a restriction flag that modifies which prediction modes are available. When the flag is set, bi-prediction is disabled for blocks of certain sizes, directly controlling memory bandwidth usage while preserving coding efficiency through selective mode application

Inventive Principle:
Principle #35Parameter changes

2Productivity

If inter-picture bi-prediction is enabled for all blocks, then coding efficiency is improved, but memory bandwidth consumption increases beyond acceptable limits

Engineering Contradiction:
Improvecoding efficiencyVSAvoidmemory bandwidth
Core Design Contradiction:
ProductivityVSQuantity of substance

Solution Approach 1:

The patent applies local quality by making prediction mode availability block-dependent. Instead of uniformly enabling or disabling bi-prediction across all blocks, the restriction flag selectively applies to specific blocks based on their characteristics, allowing coding efficiency to be maintained where needed while limiting memory bandwidth consumption where appropriate

Inventive Principle:
Principle #3Local quality

3Manufacturing precision

If prediction block size is reduced to improve detail quality, then video fidelity is improved, but memory bandwidth requirements increase due to more blocks

Engineering Contradiction:
Improvevideo fidelityVSAvoidmemory bandwidth
Core Design Contradiction:
Manufacturing precisionVSQuantity of substance

Solution Approach 1:

The patent introduces dynamics by making the prediction mode flexible and adaptive to block size. The restriction flag dynamically controls which prediction modes are available based on block characteristics, allowing the system to optimize between video fidelity and memory bandwidth consumption rather than using a fixed approach

Inventive Principle:
Principle #15Dynamics

Data Source

PatentEP3158737B1Systems and methods for intra-block copy
Publication Date: 2020.04.22 QUALCOMM INC
  • EP3158737B1 patent drawingFigure 1~2
  • EP3158737B1 patent drawingFigure 3~4
  • EP3158737B1 patent drawingFigure 5

AI summary

Techniques and systems are provided for encoding and decoding video data. For example, a method of encoding video data includes obtaining video data at an encoder, and determining to perform intra-picture prediction on the video data, using intra-block copy prediction, to generate the plurality of encoded video pictures. The method also includes performing the intra-picture prediction on the video data using the intra-block copy prediction, and, in response to determining to perform the intra-picture prediction on the video data using the intra-block copy prediction, disabling at least one of inter-picture bi-prediction or inter-picture uni-prediction for the plurality of encoded video pictures. The method also includes generating the plurality of encoded video pictures based on the received video data according to the performed intra-block copy prediction.