Constrained Disparity Vector Derivation in 3D Video Coding

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current 3D video coding systems face inefficiencies in processing and storage due to the large number of inter-view candidates required for disparity vector derivation, which increases processing time and storage needs without significantly impacting system performance.

Innovation Solution

The method involves conditionally applying constraints to disparity vectors (DVs) to reduce the vertical or horizontal components of DVs, allowing for unified application across selected coding tools, excluding disparity compensated prediction (DCP), and using 1D or 2D filters for interpolation based on DV constraint settings.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If multiple cameras are used to generate multi-view video with a large number of video sequences, then the viewing experience and realism are improved, but the storage space and transmission bandwidth requirements increase significantly

Engineering Contradiction:
Improveviewing experienceVSAvoidstorage space
Core Design Contradiction:
Adaptability or versatilityVSQuantity of substance

Solution Approach 1:

The patent merges multiple view video sequences by exploiting inter-view redundancy and correlation. Instead of storing each view independently, the system combines them into a unified structure where adjacent views share common motion vectors and disparity information, significantly reducing the total storage requirement while maintaining the ability to present multiple viewpoints.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The patent creates a universal coding framework that handles multiple views through a single set of tools. The same motion vector and disparity vector derivation mechanisms work across all views, allowing the system to efficiently encode and decode multi-view video without requiring view-specific processing, thus reducing overall resource consumption.

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

2Ease of manufacture

If conventional video coding techniques are applied to each single-view video sequence independently, then the coding simplicity is maintained, but the coding efficiency deteriorates significantly

Engineering Contradiction:
Improvecoding simplicityVSAvoidcoding efficiency
Core Design Contradiction:
Ease of manufactureVSProductivity

Solution Approach 1:

The patent combines multiple coding tools (motion-compensated prediction, disparity-compensated prediction, and inter-view residual prediction) into a unified framework that processes multiple views simultaneously. This merged approach maintains relative simplicity while dramatically improving efficiency by leveraging inter-view correlations that single-view coding cannot exploit.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The patent performs preliminary derivation of motion vectors and disparity vectors from reference views before coding the current view. By pre-computing these parameters and storing them in candidate lists, the system prepares reusable prediction data that speeds up the actual coding process and improves efficiency without adding significant complexity.

Inventive Principle:
Principle #10Preliminary action

3Productivity

If inter-view redundancy is exploited to improve coding efficiency, then the compression performance is improved, but the processing complexity and time increase

Engineering Contradiction:
Improvecompression efficiencyVSAvoidprocessing time
Core Design Contradiction:
ProductivityVSLoss of time

Solution Approach 1:

The patent performs preliminary derivation of motion vectors and disparity vectors from reference views and stores them in candidate lists. This pre-computation allows the actual coding process to simply retrieve and apply pre-derived parameters rather than computing them in real-time, significantly reducing processing time while maintaining high compression efficiency through inter-view redundancy exploitation.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent derives motion vectors and disparity vectors for multiple candidate blocks even though not all candidates will be used in the final coding decision. By preparing excess candidate data upfront, the system enables efficient runtime selection and prediction without needing to perform complex computations during the actual coding process, thus reducing overall processing time.

Inventive Principle:
Principle #16Partial or excessive action

4Quantity of substance

If disparity vectors are constrained to reduce vertical or horizontal components, then the memory usage and bitrate are reduced, but the performance may degrade for unrectified views

Engineering Contradiction:
Improvememory usageVSAvoidcoding performance
Core Design Contradiction:
Quantity of substanceVSReliability

Solution Approach 1:

The patent implements dynamic constraint application where the degree of disparity vector constraining is adjusted based on view characteristics. For rectified views, strong constraints are applied to reduce memory usage and bitrate; for unrectified views, the constraints are relaxed or adapted to maintain coding performance. This dynamic adaptation allows the system to optimize memory efficiency without sacrificing reliability when needed.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentUS9998760B2Method and apparatus of constrained disparity vector derivation in 3D video coding
Publication Date: 2018.06.12 HFI INNOVATION INC

AI summary

A method and apparatus for three-dimensional video encoding or decoding with conditionally constrained disparity vector are disclosed. In one embodiment, a derived DV (disparity vector) for the current texture block is determined and DV constraint is applied or is not applied to the derived DV to obtain a final derived DV. Inter-view predictive encoding or decoding is then applied to the input data utilizing at least one of selected coding tools, wherein a same final derived DV is used by all selected coding tools, and the selected coding tools comprise inter-view residual prediction, view synthesis prediction and inter-view motion parameter prediction.