LUT-Based Motion Vector Prediction for Video Coding Efficiency

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current video coding standards face challenges in efficiently encoding and decoding digital video due to high bandwidth demands and limitations in motion information prediction, particularly when neighboring blocks have different motion trajectories.

Innovation Solution

The implementation of Look-Up Table (LUT)-based motion vector prediction techniques, which maintain tables of motion candidates and their associated information to enhance coding efficiency by using historical data for predicting motion information, allowing for more accurate and efficient encoding and decoding processes.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional motion prediction using neighboring blocks is used, then encoding simplicity is maintained, but coding efficiency deteriorates when neighboring blocks have different motion trajectories

Engineering Contradiction:
Improvemotion prediction accuracyVSAvoidprediction structure complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent maintains a history-based motion vector prediction table that stores motion information from previously coded blocks before the current block is encoded. This preliminary storage of historical motion data enables more accurate motion prediction for the current block by selecting the best matching historical candidate, rather than relying solely on spatially neighboring blocks that may have different motion trajectories.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent copies motion information from historically coded blocks into a prediction table, creating reusable motion candidates. These copied motion vectors from historical blocks serve as predictors for the current block, allowing the system to leverage past motion patterns without requiring complex real-time calculations during the current encoding process.

Inventive Principle:
Principle #26Copying

2Measurement precision

If more motion candidates are stored in prediction tables, then prediction accuracy improves, but memory requirements and processing complexity increase

Engineering Contradiction:
Improvemotion candidate prediction accuracyVSAvoidmemory storage requirements
Core Design Contradiction:
Measurement precisionVSQuantity of substance

Solution Approach 1:

The patent implements a limited-size history-based motion vector prediction table that stores only the most relevant motion candidates from previously coded blocks. Rather than storing all historical motion data, the system maintains a selective subset of motion candidates that are most likely to be useful predictors, optimizing the balance between prediction accuracy and memory usage.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent manages the prediction table by discarding older or less relevant motion candidates when the table reaches its capacity, and recovering/replacing them with new motion candidates from recently coded blocks. This dynamic management ensures the table contains the most current and relevant motion information while maintaining a bounded memory footprint.

Inventive Principle:
Principle #34Discarding and recovering

Data Source

PatentUS12167018B2Interaction between LUT and AMVP
Publication Date: 2024.12.10 BYTEDANCE INC
  • US12167018B2 patent drawing
  • US12167018B2 patent drawing
  • US12167018B2 patent drawing

AI summary

A method of video decoding is provided to include maintaining tables, wherein each table includes a set of motion candidates and each motion candidate is associated with corresponding motion information; and performing a conversion between a first video block and a bitstream representation of a video including the first video block, the performing of the conversion including using at least some of the set of motion candidates as a predictor to process motion information of the first video block.