V-Mesh Bitstream Syntax for 3D Textured Mesh Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video-based coding standards like ISO/IEC 23090-5 Visual Volumetric Video-Based Coding (V3C) do not support the efficient compression of 3D textured static and dynamic meshes, which are crucial for representing 3D volumetric data such as point clouds and 3DoF+ content.
Innovation Solution
A bitstream structure and decoding process are developed for video-based dynamic mesh codecs (v-mesh) that incorporate new and repurposed syntax elements from V3C to efficiently encode and decode 3D textured static and dynamic meshes, utilizing techniques like pre-processing, wavelet transforms, and motion compression to enhance compression efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If video-based coding standards (V3C) are used to compress 3D data, then compression efficiency is improved, but support for 3D textured static and dynamic meshes is lacking
Solution Approach 1:
The patent extends the V3C video-based coding framework to handle multiple 3D data types including point clouds, 3DoF content, and 3D textured static/dynamic meshes. By making the coding structure universal and adaptable to different mesh types through configurable syntax elements and processing modes, the system achieves both compression efficiency and broad 3D content support
2Adaptability or versatility
If new syntax elements are introduced for mesh coding, then mesh compression capability is improved, but bitstream structure complexity increases
Solution Approach 1:
The patent nests mesh-specific syntax elements within the existing V3C bitstream structure. New elements like mesh configuration parameters, attribute metadata, and reconstruction instructions are organized in hierarchical layers that integrate with the standard video coding framework, allowing mesh functionality to be embedded without creating a completely separate complex structure
Solution Approach 2:
The patent introduces preliminary configuration syntax elements that define mesh coding parameters before the actual mesh data is encoded. These pre-configured parameters establish the framework for subsequent mesh-specific operations, reducing the complexity of inline mesh coding by preparing the bitstream structure in advance
3Manufacturing precision
If 3D textured meshes are represented with high detail, then representation quality is improved, but data size and compression difficulty increase
Solution Approach 1:
The patent segments 3D textured meshes into discrete components including geometry data, texture coordinate data, and attribute data. Each segment can be independently compressed using optimized video coding techniques, allowing high-detail representation to be achieved while managing data size through selective compression of different mesh components
Solution Approach 2:
The patent projects 3D mesh data onto 2D planes for video-based compression, then uses metadata to reconstruct the 3D representation. This dimensionality transformation allows leveraging efficient 2D video coding algorithms while maintaining 3D quality, effectively reducing data size without sacrificing representation fidelity
Data Source
AI summary
A video dynamic mesh coding (v-DMC) decoding system, includes a de-multiplexer that receives and demultiplexes an encoded v-DMC bitstream into: a parameter set and mesh data, geometry, atlas data, and attribute video substreams. The decoding system also includes: a mesh data substream decoder; a video decoder that decodes the geometry data substream; an atlas data substream decoder; a mesh subdivision component that subdivides the one or more base meshes into one or more resampled base meshes based upon the decoded atlas data; a displacement decoder that outputs one or more displacements to verticies of the one or more resampled base meshes; a mesh position refinement component that applies the one or more displacements to the one or more resampled base meshes and outputs one or more resultant meshes; and a video decoder that decodes the attribute video substream into one or more texture images.