Motion Information Encoding Candidate Generation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional video codecs like MPEG-4 H.264/AVC face challenges in achieving high coding efficiency for high-definition video content due to the significant contribution of motion vectors to the encoded bitstream, necessitating effective methods to reduce the transmission overhead of motion vector information.

Innovation Solution

The method generates various candidate motion information by utilizing motion information from spatially and temporally collocated prediction units, allowing for the selection of optimal motion vectors to enhance compression efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If motion vectors are transmitted for each block to achieve accurate motion compensation, then compression accuracy is improved, but transmission overhead increases

Engineering Contradiction:
Improvecompression accuracyVSAvoidtransmission overhead
Core Design Contradiction:
Measurement precisionVSLoss of information

Solution Approach 1:

The patent applies preliminary action by generating candidate motion vectors in advance using motion information from spatially and temporally collocated prediction units before the actual encoding process. This pre-computation allows the encoder to select from pre-prepared candidates, reducing the need to transmit full motion vector information while maintaining accurate motion compensation through selective candidate usage.

Inventive Principle:
Principle #10Preliminary action

2Productivity

If motion information from multiple prediction units is utilized to generate candidate motion vectors, then compression efficiency is improved, but encoding complexity increases

Engineering Contradiction:
Improvecompression efficiencyVSAvoidencoding complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent segments the motion information source into distinct spatial and temporal components. It separately processes motion information from spatially collocated prediction units (adjacent blocks in the current picture) and temporally collocated prediction units (corresponding blocks in reference pictures), then combines these segmented results to form candidate motion vectors. This segmentation approach systematically manages complexity while improving compression efficiency.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent creates a universal candidate motion vector generation mechanism that serves multiple prediction units simultaneously. The same set of candidate motion vectors derived from spatial and temporal collocated units can be used across different prediction units, making the encoding process more efficient and reducing overall complexity through reusable candidate sets.

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

Data Source

PatentEP3876540B1Method for decoding and encoding motion information, apparatus for encoding motion information and computer-readable medium
Publication Date: 2023.06.07 SAMSUNG ELECTRONICS CO LTD
  • EP3876540B1 patent drawingFigure 1~2
  • EP3876540B1 patent drawingFigure 3
  • EP3876540B1 patent drawingFigure 4

AI summary

A method and apparatus for encoding and decoding motion information. The encoding method includes determining whether motion information of spatial prediction units that are spatially collocated to a current prediction unit and motion information of temporal prediction units that are temporally collocated to the current prediction are available; when the number of the motion information of the spatial prediction units and the motion information of the temporal prediction units is less than a predetermined number, generating additional candidate motion information by using the available motion information of the spatial prediction units and the motion information of the temporal prediction units such that the total number of pieces of candidate motion information is the predetermined number; and decoding motion information of the current prediction unit by using the 'n' pieces of motion information.