Video Block Triangular Motion Compensation for Non-Aligned Edges

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing video encoding methods face poor prediction effects and large coding residuals when dealing with blocks at the edges of objects that are not horizontally or vertically aligned, leading to poor coding performance.

Innovation Solution

Divide current blocks into first and second triangle sub-blocks with different target motion information for improved prediction accuracy and coding performance by using geometric partitioning with triangular partitions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If a single prediction mode is used for the current rectangular block, then the encoding process is simple, but the prediction accuracy deteriorates and coding residuals increase

Engineering Contradiction:
Improveencoding process complexityVSAvoidprediction accuracy
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The current block is divided into multiple triangular prediction units (first triangular PU and second triangular PU) with different partitioning orientations. Each triangular PU can independently select from motion vector candidate lists, allowing different prediction modes for different regions of the same block. This segmentation enables the encoder to adapt to non-uniform motion patterns within a block, improving prediction accuracy without requiring a completely complex encoding framework.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Different triangular prediction units are assigned different motion vector candidate lists based on their specific characteristics and requirements. The first triangular PU uses a first motion vector candidate list while the second triangular PU uses a second motion vector candidate list. This local differentiation allows each region to receive optimized prediction parameters tailored to its specific motion characteristics, thereby improving overall prediction accuracy while maintaining manageable encoding complexity through localized optimization.

Inventive Principle:
Principle #3Local quality

2Measurement precision

If triangular prediction units with different motion information are used, then prediction accuracy improves, but the encoding and decoding complexity increases

Engineering Contradiction:
Improveprediction accuracyVSAvoidencoding and decoding complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system dynamically selects motion vector candidate lists for different triangular prediction units based on their partitioning orientations and characteristics. The encoder determines which motion vector candidate list to use for each triangular PU, and this selection is signaled to the decoder. This dynamic adaptation allows the system to optimize prediction accuracy for different regions while managing complexity through structured decision-making processes that are efficiently implemented in both encoding and decoding.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The system changes the parameter of motion vector candidate list selection based on the partitioning orientation of each triangular prediction unit. By varying this parameter according to the specific characteristics of each triangular PU, the system achieves optimized prediction for different regions. The parameter change is implemented through clear signaling mechanisms that allow the decoder to reconstruct the same prediction structure, balancing improved accuracy with manageable complexity through systematic parameter variation.

Inventive Principle:
Principle #35Parameter changes

3Productivity

If motion vector candidate lists are selected for different triangular sub-blocks, then coding performance improves, but the signaling and processing requirements increase

Engineering Contradiction:
Improvecoding performanceVSAvoidsignaling and processing requirements
Core Design Contradiction:
ProductivityVSEase of operation

Solution Approach 1:

The system performs preliminary actions by pre-defining multiple motion vector candidate lists with different characteristics before the actual encoding process. These candidate lists are prepared in advance and can be selectively assigned to different triangular prediction units. This preliminary preparation allows the encoder to quickly make optimal selections during encoding without requiring complex real-time calculations, thereby improving coding performance while reducing the computational burden during the actual encoding and decoding operations.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system uses copying by replicating motion vector candidate lists with different characteristics for different triangular prediction units. Instead of creating entirely new motion information from scratch for each region, the system copies and adapts existing motion vector candidate lists to match the specific needs of each triangular PU. This copying approach maintains consistency with overall motion patterns while allowing regional optimization, improving coding performance with relatively simple processing requirements since it builds upon existing structures rather than creating new ones.

Inventive Principle:
Principle #26Copying

Data Source

PatentEP3989573B1Decoding method and device
Publication Date: 2025.08.13 HANGZHOU HIKVISION DIGITAL TECHNOLOGY CO LTD
  • EP3989573B1 patent drawingFigure 1~2
  • EP3989573B1 patent drawingFigure 3
  • EP3989573B1 patent drawingFigure 4~5B

AI summary

The present application provides an encoding and decoding method, apparatus and device, the method comprises: dividing, if characteristic information of a current block meets specific conditions, the current block into a first triangle sub-block and a second triangle sub-block; constructing a motion information list for the current block, the motion information list includes multiple pieces of candidate motion information; acquiring first target motion information of the first triangle sub-block and second target motion information of the second triangle sub-block from the motion information list; the first target motion information is different from the second target motion information; performing motion compensation on the first triangle sub-block based on the first target motion information to obtain a prediction value of the first triangle sub-block; performing motion compensation on the second triangle sub-block based on the second target motion information to obtain a prediction value of the second triangle sub-block. Through the technical solution of the present application, the ending performance can be improved.