Video Encoding Prediction Information Reduction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In video encoding technology, the increased number of partitioned coding units in a quadtree structure leads to an increase in prediction information quantity, resulting in a decreased compression ratio.

Innovation Solution

The method involves selecting either the prediction processing of a parent CU or utilizing its prediction image for encoding target CUs, storing flag information to indicate this selection, and omitting prediction processing for CUs with low prediction accuracy, thereby reducing the information quantity required for encoding.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If the number of partitioned coding units in quadtree structure is increased to improve prediction accuracy, then prediction accuracy is improved, but the coding quantity of prediction information increases and compression ratio drops

Engineering Contradiction:
Improveprediction accuracyVSAvoidcompression ratio
Core Design Contradiction:
Measurement precisionVSLoss of information

Solution Approach 1:

The patent applies segmentation by dividing the macroblock into variable-sized coding units (16×16, 8×8, 4×4, 2×2) based on picture characteristics. This allows the system to segment the image into appropriate blocks for prediction processing, improving prediction accuracy while controlling the amount of prediction information needed by adapting the segmentation level to local image content rather than uniformly partitioning the entire macroblock.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent implements dynamics by enabling flexible switching between different prediction modes (intra-prediction, inter-prediction, skip mode) and different coding unit sizes based on local picture characteristics. The system dynamically selects the optimal prediction mode and block size for each coding unit, allowing adaptation to varying image content and motion patterns, thereby improving prediction accuracy without uniformly increasing prediction information across the entire macroblock.

Inventive Principle:
Principle #15Dynamics

2Measurement precision

If prediction processing is performed on each partitioned coding unit, then prediction accuracy is improved, but the information quantity for describing prediction processing increases

Engineering Contradiction:
Improveprediction accuracyVSAvoidprediction information quantity
Core Design Contradiction:
Measurement precisionVSQuantity of substance

Solution Approach 1:

The patent applies copying by utilizing reference blocks from previously decoded neighboring blocks to generate prediction values for current coding units. Instead of performing complex prediction processing on every partitioned coding unit, the system copies pixel values from reference blocks (e.g., using copy mode in intra-prediction or motion compensation in inter-prediction), thereby achieving good prediction accuracy with reduced prediction information quantity.

Inventive Principle:
Principle #26Copying

Solution Approach 2:

The patent extracts and utilizes only the essential prediction information needed for each coding unit by selecting from multiple prediction modes. The system extracts reference block data and motion vector information only where necessary, rather than processing all coding units with full prediction algorithms, thereby reducing the overall prediction information quantity while maintaining accuracy where needed.

Inventive Principle:
Principle #2Taking out (Extraction)

Data Source

PatentEP2736254B1Video decoding method and image encoding method
Publication Date: 2018.07.04 HITACHI LTD
  • EP2736254B1 patent drawingFigure 1
  • EP2736254B1 patent drawingFigure 2
  • EP2736254B1 patent drawingFigure 3

AI summary

A video decoding method and image encoding method capable of utilizing a portion of the prediction image predicted in a size larger than the size of the macroblock of the encoding target, as the prediction image of the macroblock of the encoding target in order to lower the compression ratio in image coding accompanied by expansion and contraction of blocks containing coding units.