Sub-Prediction Unit Motion Vector Signaling for Video Decoding

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing video coding techniques face inefficiencies in motion vector prediction, particularly in deriving accurate predictors from neighboring blocks, which affects compression efficiency and decoding performance.

Innovation Solution

The proposed method involves adaptively selecting a motion vector predictor from lists of sub-prediction unit and prediction unit level candidates, utilizing spatial and temporal neighbors, and employing techniques like affine motion vector prediction, alternative temporal motion vector prediction, spatial-temporal motion vector prediction, planar motion vector prediction, and pattern matched motion vector derivation to enhance prediction accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional motion vector prediction is used, then decoding complexity is reduced, but prediction accuracy deteriorates

Engineering Contradiction:

Improvemotion vector prediction accuracyVSAvoiddecoding complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The prediction unit is divided into multiple sub-prediction units, and motion vector prediction is performed separately for each sub-prediction unit. This segmentation allows for more accurate local motion representation while maintaining manageable decoding complexity through structured candidate list management.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent dynamically adjusts the motion prediction granularity by selecting between sub-PU level and PU level prediction based on a flag in the encoded data. This dynamic adaptation allows the system to optimize between accuracy and complexity on a block-by-block basis.

Inventive Principle:

Principle #15Dynamics

2Measurement precision

If sub-prediction unit level motion prediction is used, then prediction accuracy is improved, but computational complexity increases

Engineering Contradiction:

Improvemotion prediction accuracyVSAvoidcomputational power consumption

Core Design Contradiction:

Measurement precisionVSPower

Solution Approach 1:

The patent applies sub-PU level motion prediction selectively rather than universally. By using a flag to indicate when sub-PU level prediction is applied, the system performs the more computationally intensive operation only where necessary, achieving improved accuracy without excessive overall computational cost.

Inventive Principle:

Principle #16Partial or excessive action

Solution Approach 2:

The patent changes the granularity parameter of motion prediction from fixed PU level to variable sub-PU level based on the prediction mode flag. This parameter change allows the system to adapt computational resources to the actual prediction needs of different video blocks.

Inventive Principle:

Principle #35Parameter changes

3Adaptability or versatility

If multiple motion prediction candidate lists are maintained, then prediction versatility is improved, but memory requirements increase

Engineering Contradiction:

Improveprediction mode versatilityVSAvoidmemory storage requirements

Core Design Contradiction:

Adaptability or versatilityVSQuantity of substance

Solution Approach 1:

The candidate lists are segmented by prediction level (sub-PU level candidates and PU level candidates) and organized in a hierarchical structure. This segmentation allows efficient memory management by storing candidates in structured groups that can be selectively accessed based on the prediction mode being used.

Inventive Principle:

Principle #1Segmentation

Data Source

PatentUS20250294181A1Signaling sub-prediction unit motion vector predictor

Publication Date: 2025.09.18 QUALCOMM INC

AI summary

A method of decoding video data comprising parsing a sub-prediction unit motion flag from received encoded video data, deriving a list of sub-prediction unit level motion prediction candidates if the sub-prediction unit motion flag is active, deriving a list of prediction unit level motion prediction candidates if the sub-prediction unit motion flag is not active, and decoding the encoded video data using a selected motion vector predictor.