3D-HEVC Motion Vector Candidate List Derivation Simplification

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current video encoding and decoding technologies face challenges in simplifying motion vector candidate derivation and adding new candidates for 3D-HEVC coding processes, particularly in merge mode and advanced motion vector prediction, which affects the efficiency of multiview video coding.

Innovation Solution

The proposed solution involves generating and utilizing motion vector candidate lists for video encoders and decoders, including additional candidates by combining partial motion vectors, modifying existing ones, or inserting zero vectors, to enhance the prediction accuracy and reduce bit requirements in 3D-HEVC coding processes.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If motion vector candidate lists are generated using traditional methods in 3D-HEVC coding, then the coding process can be completed with standard procedures, but the prediction accuracy is insufficient and bit usage is excessive

Engineering Contradiction:
Improveprediction accuracyVSAvoidbit usage
Core Design Contradiction:
Measurement precisionVSQuantity of substance

Solution Approach 1:

The motion vector candidate derivation process is segmented into multiple stages: initial candidate list generation, candidate refinement based on view synthesis prediction, and selective insertion of additional candidates. This segmentation allows the system to achieve high prediction accuracy by focusing computational effort on the most promising candidates while avoiding the generation of excessive bit data through systematic filtering and selection processes.

Inventive Principle:
Principle #1Segmentation

2Productivity

If the motion vector candidate derivation process is simplified, then the encoding and decoding efficiency improves, but the prediction accuracy may be compromised

Engineering Contradiction:
Improveencoding and decoding efficiencyVSAvoidprediction accuracy
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The system performs preliminary actions by pre-generating a comprehensive set of motion vector candidates using view synthesis prediction techniques before the actual encoding process. This preliminary candidate generation includes inserting additional candidates derived from reference view pictures, which establishes a robust foundation that maintains high prediction accuracy while allowing the subsequent simplified derivation process to operate efficiently without compromising quality.

Inventive Principle:
Principle #10Preliminary action

3Measurement precision

If additional motion vector candidates are added to the candidate list, then prediction accuracy improves, but the complexity of the candidate derivation process increases

Engineering Contradiction:
Improveprediction accuracyVSAvoidcandidate derivation process complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system applies local quality by adding additional motion vector candidates specifically at positions in the candidate list where they are most beneficial for prediction accuracy. Rather than uniformly complicating the entire derivation process, the method selectively inserts candidates derived from view synthesis prediction at specific locations in the merge candidate list, thereby improving prediction accuracy locally where it matters most while keeping the overall process complexity manageable through targeted rather than comprehensive modifications.

Inventive Principle:
Principle #3Local quality

Data Source

PatentEP3155812B1Simplified shifting merge candidate and merge list derivation in 3d-hevc
Publication Date: 2023.04.05 QUALCOMM INC
  • EP3155812B1 patent drawingFigure 1
  • EP3155812B1 patent drawingFigure 2
  • EP3155812B1 patent drawingFigure 3

AI summary

A device for coding three-dimensional includes a video coder configured to add one or more motion vector candidates to a motion vector candidate list; identify, from the one or more motion vector candidates in the motion vector candidate list, a disparity motion vector candidate; based on the disparity motion vector candidate, determine a disparity shifted motion vector (DSMV)candidate; and add the DSMV candidate to the motion vector candidate list; and code the block of video data using a motion vector corresponding to the identified candidate.