Multi-View Video Depth Coding with Parametric Reconstruction

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing immersive video coding methods require intensive computational resources and result in suboptimal image quality due to the need for depth estimation on lesser-quality decoded texture components, which are not efficiently coded or transmitted.

Innovation Solution

A method for coding and decoding 3D scene views that partitions depth components into blocks, obtains depth information from texture data, and codes depth estimation parameters instead of the depth blocks themselves, reducing computational load and memory requirements.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If depth maps are coded and transmitted in conventional immersive video coding, then image quality is improved, but bit rate increases significantly

Engineering Contradiction:

Improveimage qualityVSAvoidbit rate

Core Design Contradiction:

Manufacturing precisionVSQuantity of substance

Solution Approach 1:

The patent extracts only the essential depth information needed for view synthesis by coding depth estimation parameters rather than complete depth maps. This selective extraction maintains sufficient image quality for synthesis while dramatically reducing the quantity of data that needs to be transmitted.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The patent changes the representation parameters from full depth maps to compressed depth estimation parameters. By transforming the data representation from detailed spatial information to parametric information, the bit rate is reduced while the essential depth characteristics are preserved for accurate view synthesis.

Inventive Principle:

Principle #35Parameter changes

2Power

If depth estimation is performed on decoded texture components, then computational resources are reduced, but image quality deteriorates due to lower texture quality

Engineering Contradiction:

Improvecomputational resourcesVSAvoidimage quality

Core Design Contradiction:

PowerVSManufacturing precision

Solution Approach 1:

The patent performs depth estimation on original, high-quality texture components before decoding and compression losses occur. By conducting the depth estimation operation in advance while the source material is still at full quality, the resulting depth maps maintain high accuracy even though the subsequent decoded textures may be of lower quality.

Inventive Principle:

Principle #10Preliminary action

3Measurement precision

If complete depth maps are coded and transmitted, then view synthesis accuracy is improved, but memory requirements increase

Engineering Contradiction:

Improveview synthesis accuracyVSAvoidmemory requirements

Core Design Contradiction:

Measurement precisionVSQuantity of substance

Solution Approach 1:

Instead of storing and transmitting complete depth map data, the patent uses depth estimation parameters that can be used to reconstruct or copy the necessary depth information at the decoder side. This parametric copying approach maintains view synthesis accuracy while requiring minimal memory storage and transmission bandwidth.

Inventive Principle:

Principle #26Copying

Data Source

PatentUS12382080B2Method for encoding and decoding a multi-view video

Publication Date: 2025.08.05 ORANGE SA

AI summary

A method for coding views simultaneously representing a 3D scene from different positions or different view angles, implemented by a coding device. The method includes, for a depth component of at least one view: partitioning the depth component into at least one block; obtaining depth information of the at least one block from texture data of a texture component of at least one of the views; obtaining at least one depth estimation parameter from the information; and coding the at least one depth estimation parameter, the depth information of the at least one block not being coded.