Disparity Vector Derivation in 3D Video Coding

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current video coding technologies face challenges in efficiently deriving disparity vectors for 3D video coding, particularly in multiview video coding, which affects the accuracy and efficiency of inter-view prediction and block-based view synthesis prediction.

Innovation Solution

The proposed solution involves a method for deriving disparity vectors by performing texture first coding for multiple views, locating corresponding depth blocks, and using depth values to calculate disparity vectors for macroblocks, allowing for efficient coding of sub-blocks based on these vectors.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If traditional disparity vector derivation methods are used in multiview video coding, then inter-view prediction can be performed, but the complexity of the coding process increases and efficiency decreases

Engineering Contradiction:
Improvecoding efficiencyVSAvoiddisparity vector derivation complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent performs texture first coding for all views before depth coding, establishing a coding order that enables efficient disparity vector derivation. By pre-coding texture information and establishing correspondence between texture and depth blocks in advance, the system eliminates the need for complex real-time disparity calculations during decoding, thereby improving coding efficiency while reducing complexity

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent divides the macroblock into sub-blocks and derives disparity vectors for each sub-block independently based on corresponding depth block values. This segmentation approach allows parallel processing and simplifies the overall derivation process by breaking down the complex macroblock-level operation into simpler, independent sub-block operations

Inventive Principle:
Principle #1Segmentation

2Measurement precision

If depth-based disparity vector derivation is implemented, then inter-view prediction accuracy improves, but the requirement for depth map processing and computational resources increases

Engineering Contradiction:
Improveinter-view prediction accuracyVSAvoiddepth processing complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent introduces depth maps as an intermediary element that facilitates accurate disparity vector derivation without requiring direct complex calculations between texture blocks of different views. By using the depth map as a mediator that encodes spatial relationship information, the system achieves high prediction accuracy while the actual computation is performed during the texture-first coding phase, reducing runtime complexity

Inventive Principle:
Principle #24Intermediary (Mediator)

3Productivity

If macroblocks are coded using derived disparity vectors, then block-based view synthesis prediction efficiency improves, but the need to process and store depth block information increases

Engineering Contradiction:
Improveview synthesis prediction efficiencyVSAvoiddepth block data volume
Core Design Contradiction:
ProductivityVSQuantity of substance

Solution Approach 1:

The patent applies disparity vector derivation selectively to sub-blocks that benefit from inter-view prediction, rather than uniformly to all macroblocks. By using the derived disparity vectors only where needed for view synthesis prediction and leveraging the already-coded texture information from the texture-first approach, the system achieves high efficiency without unnecessarily processing and storing depth block data for all regions

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP2976884B1Simplifications on disparity vector derivation and motion vector prediction in 3D video coding
Publication Date: 2020.03.11 QUALCOMM INC
  • EP2976884B1 patent drawingFigure 1
  • EP2976884B1 patent drawingFigure 2
  • EP2976884B1 patent drawingFigure 3

AI summary

A video coder can be configured to perform texture first coding for a first texture view, a first depth view, a second texture view, and a second depth view; for a macroblock of the second texture view, locate a depth block of the first depth view that corresponds to the macroblock; based on at least one depth value of the depth block, derive a disparity vector for the macroblock; code a first sub-block of the macroblock based on the derived disparity vector; and, code a second sub-block of the macroblock based on the derived disparity vector.