Asymmetric Motion Partitioning in 3D Video Coding

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current video coding techniques, such as those in 3D-HEVC, face challenges in efficiently encoding and decoding video data, particularly in partitioning prediction units and deriving motion vectors for asymmetric motion partitioning, which affects coding efficiency and compression performance.

Innovation Solution

The proposed solution involves using backward view synthesis prediction (BVSP) with asymmetric motion partitioning, where video blocks are partitioned into sub-blocks of sizes 8x4 or 4x8, and disparity motion vectors are derived from corresponding depth blocks to synthesize reference blocks for motion compensation, enhancing encoding and decoding efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If asymmetric motion partitioning is used to improve coding efficiency, then compression performance is improved, but the complexity of deriving motion vectors and partitioning prediction units increases

Engineering Contradiction:
Improvecoding efficiencyVSAvoidcomplexity of deriving motion vectors
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The prediction unit is divided into sub-blocks of different sizes (e.g., 8x4 or 4x8) to allow asymmetric motion partitioning. This segmentation enables more flexible motion representation while managing complexity through structured subdivision of the coding unit

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies asymmetric motion partitioning by creating sub-blocks with non-uniform dimensions (such as 8x4 or 4x8 instead of uniform 8x8). This asymmetry allows better adaptation to directional motion patterns, improving compression efficiency by matching the actual motion characteristics in the video content

Inventive Principle:
Principle #4Asymmetry

2Productivity

If backward view synthesis prediction is used to synthesize reference blocks, then compression performance is improved, but the computational complexity of deriving disparity motion vectors increases

Engineering Contradiction:
Improvecompression performanceVSAvoidcomputational complexity of deriving disparity motion vectors
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

Depth information is derived and stored in advance for use in backward view synthesis prediction. By preparing depth maps beforehand, the system can efficiently synthesize reference blocks during decoding without performing complex depth estimation at that stage, thus improving compression performance while managing computational complexity

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

Depth maps serve as an intermediary element between the current picture and reference pictures. The depth information mediates the synthesis process by providing geometric cues that simplify the derivation of disparity motion vectors, reducing direct computational complexity while maintaining compression efficiency

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentEP3044961B1Video coding techniques using asymmetric motion partitioning
Publication Date: 2020.03.18 QUALCOMM INC
  • EP3044961B1 patent drawingFigure 1
  • EP3044961B1 patent drawingFigure 2
  • EP3044961B1 patent drawingFigure 3

AI summary

Techniques for decoding video data include receiving residual data corresponding to a block of video data, wherein the block of video data is encoded using asymmetric motion partitioning, is uni-directionally predicted using backward view synthesis prediction (BVSP), and has a size of 16x12, 12x16, 16x4 or 4x16, partitioning the block of video data into sub-blocks, each sub-block having a size of 8x4 or 4x8, deriving a disparity motion vector for each of the sub-blocks from a corresponding depth block in a depth picture corresponding to a reference picture, synthesizing a respective reference block for each of the sub-blocks using the respective derived disparity motion vector, and decoding the block of video data by performing motion compensation on each of the sub-blocks using the residual data and the synthesized respective reference blocks.