Motion Vector Precision Rounding for Video Encoding

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current video compression and decompression techniques face challenges in efficiently managing motion vector precisions across different neighboring blocks, leading to suboptimal candidate list generation and increased bandwidth usage, especially when dealing with high-quality video demands.

Innovation Solution

The proposed solution involves an encoder that rounds motion vector precisions uniformly and generates candidate lists by considering the precision of the current block and neighboring blocks, applying specific pseudo code rules to determine the best candidate motion vectors, and signaling these precisions at various levels such as slice, region, or tile, to optimize coding efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If different motion vector precisions are used for different neighboring blocks, then the candidate list can be more adaptive to local motion characteristics, but the complexity of precision management and candidate list generation increases

Engineering Contradiction:
Improveadaptability to local motion characteristicsVSAvoidcomplexity of precision management
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent applies local quality by allowing different motion vector precisions (e.g., 1/4-pel, 1/8-pel, 1/16-pel) to be used for different neighboring blocks based on their local motion characteristics. Each neighboring block can have its precision independently selected and signaled, enabling the encoder to adapt to varying motion complexities in different regions while maintaining manageable complexity through systematic precision selection rules.

Inventive Principle:
Principle #3Local quality

2Measurement precision

If high-precision motion vectors are used, then video quality improves, but bandwidth usage increases

Engineering Contradiction:
Improvemotion vector precisionVSAvoidbandwidth usage
Core Design Contradiction:
Measurement precisionVSQuantity of substance

Solution Approach 1:

The patent implements parameter changes by dynamically adjusting motion vector precision levels based on local motion activity and block characteristics. The encoder selectively uses higher precision (1/16-pel) only when necessary for accurate motion representation, while using lower precision (1/4-pel or 1/8-pel) for regions with simpler motion patterns. This adaptive parameter adjustment maintains high video quality where needed while reducing overall bandwidth consumption.

Inventive Principle:
Principle #35Parameter changes

3Measurement precision

If motion vector precisions are not uniformly rounded, then more precise candidate selection is possible, but the complexity of candidate list generation and processing increases

Engineering Contradiction:
Improvecandidate selection precisionVSAvoidcomplexity of candidate list generation
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent applies preliminary action by performing uniform rounding of motion vector precisions to a common reference level (e.g., rounding all precisions to 1/4-pel or 1/8-pel) before generating the candidate list. This preliminary rounding step simplifies subsequent candidate list generation and processing by eliminating the need to handle multiple precision levels simultaneously, while still maintaining adequate precision for accurate motion compensation.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP3639520B1Selection and signaling of motion vector (MV) precisions
Publication Date: 2024.02.14 HUAWEI TECH CO LTD
  • EP3639520B1 patent drawingFigure 1
  • EP3639520B1 patent drawingFigure 2
  • EP3639520B1 patent drawingFigure 3

AI summary

An apparatus comprises: a memory; and a processor coupled to the memory and configured to: obtain a current block of a video frame, obtain candidate MVs corresponding to neighboring blocks of the video frame, the neighboring blocks neighbor the current block, obtain precisions of the candidate MVs, perform first rounding of the precisions based on a rounding scheme, perform second rounding of the candidate MVs based on the first rounding, perform pruning of the candidate MVs, and generate a candidate list based on the second rounding and the pruning.