HMVP List Management for Sub-Block Motion Vector Prediction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current video compression systems fail to utilize motion vectors from sub-block-based inter-coded blocks for encoding and decoding, limiting the compression efficiency when a block is surrounded by sub-block-based inter-coded blocks.

Innovation Solution

A method and apparatus that store and utilize motion information from sub-block-based inter-coded blocks to enhance motion vector prediction for encoding and decoding, by adding the motion information to a list that can be used for subsequent blocks, thereby improving compression efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If only non-sub-block-based inter-coded blocks contribute to the HMVP list, then the HMVP list remains simple to manage, but blocks surrounded by sub-block-based inter-coded blocks cannot benefit from motion vectors of previously encoded/decoded blocks

Engineering Contradiction:
Improvemotion vector prediction availabilityVSAvoidHMVP list management complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent segments the motion information data into different types (first motion information data for non-sub-block-based blocks and second motion information data for sub-block-based blocks). This segmentation allows the system to selectively add appropriate motion information to the HMVP list, expanding prediction availability while maintaining organized management through distinct data categories.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The HMVP list is enhanced to serve multiple functions by accepting both first motion information data from non-sub-block-based blocks and second motion information data from sub-block-based blocks. This multi-functionality allows the same data structure to support diverse block types, improving versatility without requiring separate management systems.

Inventive Principle:
Principle #6Universality (Multi-functionality)

2Productivity

If motion information from sub-block-based inter-coded blocks is added to the HMVP list, then compression efficiency improves, but the complexity of managing motion information data increases

Engineering Contradiction:
Improvecompression efficiencyVSAvoidmotion information data management complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent applies local quality by treating different motion information data types differently in the HMVP list management process. First motion information data and second motion information data are handled with distinct processing rules and selection criteria, allowing optimized compression efficiency for each block type while maintaining systematic management through type-specific procedures.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The system changes parameters by introducing type differentiation for motion information data. By categorizing data as first motion information data or second motion information data based on block type, the system enables parameter-based selection and management strategies that improve compression efficiency while maintaining manageable complexity through structured parameter handling.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentUS20220264147A1Hmvc for affine and sbtmvp motion vector prediction modes
Publication Date: 2022.08.18 INTERDIGITAL CE PATENT HOLDINGS SAS
  • US20220264147A1 patent drawing
  • US20220264147A1 patent drawing
  • US20220264147A1 patent drawing

AI summary

An apparatus for encoding or decoding blocks of a current picture encodes sub-blocks of a first block of a current picture. The sub-blocks encoded or decoded based on motion vectors determined according a motion information data associated with the first block. In a second step, a set of second motion information data is determined as a function of the first motion information data. These second motion information data are added to a list of motion information data, that is used to determine a motion information data for further blocks of the current picture to encode or decode.