Sub-Block Temporal Motion Vector Prediction for Image Compression

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image compression technologies struggle to effectively consider various candidates for deriving a sub-block based temporal motion vector predictor, which limits the accuracy of prediction and compression performance.

Innovation Solution

An image encoding/decoding method and device that derive a temporal vector of a current block based on a motion vector of one candidate in a candidate group, determine a collocated block in a collocated picture, and perform inter prediction using a motion vector of the current block in a sub-block unit.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a limited number of candidates are used for deriving temporal motion vector predictor, then device complexity is reduced, but prediction accuracy deteriorates

Engineering Contradiction:
Improveprediction accuracyVSAvoiddevice complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent divides the current block into multiple sub-blocks and derives motion vectors for each sub-block independently using template matching. This segmentation allows the system to consider multiple candidates across different sub-blocks, improving prediction accuracy without requiring a single complex candidate selection process for the entire block.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent performs template matching on multiple sub-blocks rather than the entire block, considering more candidates than traditional methods. By applying partial action to sub-blocks and combining results, the system achieves higher prediction accuracy through excessive candidate consideration while managing complexity through localized processing.

Inventive Principle:
Principle #16Partial or excessive action

2Measurement precision

If more various candidates are considered for sub-block based temporal motion vector predictor, then prediction accuracy is improved, but computational complexity increases

Engineering Contradiction:
Improveprediction accuracyVSAvoidcomputational complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the current block into multiple sub-blocks and performs independent template matching on each. This segmentation enables consideration of multiple candidates across sub-blocks, improving prediction accuracy while distributing computational complexity across smaller, manageable units rather than processing all candidates for the entire block at once.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies template matching locally to each sub-block rather than uniformly to the entire block. This local quality approach allows the system to consider various candidates in each local region independently, improving prediction accuracy for motion-varying regions while reducing overall computational complexity by limiting the search scope to local sub-blocks.

Inventive Principle:
Principle #3Local quality

3Productivity

If sub-block based temporal motion vector predictor is implemented, then compression performance is improved, but processing complexity increases

Engineering Contradiction:
Improvecompression performanceVSAvoidprocessing complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent divides the current block into multiple sub-blocks and derives motion vectors for each sub-block using template matching against candidate blocks. This segmentation enables better adaptation to local motion variations, improving compression performance through more accurate prediction while managing processing complexity by applying the method to smaller sub-blocks rather than the entire block.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent performs template matching and candidate selection for multiple sub-blocks, considering more candidates than traditional block-based methods. This partial action applied to each sub-block improves compression performance through excessive candidate consideration, while the localized scope of each sub-block processing keeps individual complexity manageable.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP4550791A1Image encoding/decoding method and device, and recording medium having bitstream stored thereon
Publication Date: 2025.05.07 LG ELECTRONICS INC
  • EP4550791A1 patent drawingFigure 1
  • EP4550791A1 patent drawingFigure 2
  • EP4550791A1 patent drawingFigure 3

AI summary

An image decoding/encoding method and device, according to the present disclosure, may: derive a temporal vector of the current block on the basis of a motion vector of one candidate in a candidate group including a plurality of candidates; on the basis of the temporal vector, determine a collocated block of the current block, in a collocated picture; on the basis of a motion vector of the collocated block, derive a motion vector of the current block by sub-block unit; and carry out inter prediction for the current block on the basis of the motion vector of the current block.