3D Mesh Encoding for Lossless Non-Manifold Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing encoding methods, such as Edgebreaker and MPEG-based solutions, fail to achieve lossless decoding of three-dimensional mesh models with non-manifold structures due to the inability to maintain the correspondence between geometric and attribute vertices during splitting and encoding processes.
Innovation Solution
The proposed method determines first indication information to assess the connection relationship between geometric and attribute vertices, and encodes non-manifold structure information, including identification and index information of duplicate vertices, to facilitate lossless decoding by maintaining the connection relationship during the encoding and decoding processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If splitting processing is performed on non-manifold structure to enable encoding, then the mesh can be processed by standard encoding methods, but the correspondence between geometric vertices and attribute vertices is lost, preventing lossless decoding
Solution Approach 1:
The patent performs preliminary identification and recording of duplicate vertices before the splitting processing. By pre-marking which vertices are duplicates and their corresponding relationships, the system preserves the correspondence information that would otherwise be lost during the splitting process, enabling subsequent lossless reconstruction.
Solution Approach 2:
The patent introduces an intermediary data structure (the duplicate vertex identification information and correspondence relationship information) that mediates between the splitting process and the decoding process. This intermediary layer maintains the mapping between geometric vertices and attribute vertices through the splitting operation, allowing accurate reconstruction without direct loss of correspondence.
2Adaptability or versatility
If duplicate vertices are generated through splitting processing, then non-manifold structures can be encoded, but the complexity of maintaining vertex correspondence increases
Solution Approach 1:
The patent segments the vertex correspondence management into distinct components: duplicate vertex identification information, correspondence relationship information, and encoding information. This segmentation allows each component to handle specific aspects of the complexity independently, making the overall system more manageable while supporting non-manifold structures.
Solution Approach 2:
The patent creates copies of vertex identification information and correspondence relationships that are stored separately from the main mesh data. These copies serve as references during decoding to reconstruct the original vertex relationships, reducing the complexity of maintaining correspondence by externalizing the mapping information.
3Productivity
If vertex splitting is performed to handle non-manifold structures, then encoding can proceed, but information about the original non-manifold structure is lost
Solution Approach 1:
The patent temporarily discards the original non-manifold structure representation during encoding by converting it to a manifold structure through splitting. However, by recording the duplicate vertex information and correspondence relationships, the system recovers the original structure information during decoding, achieving both encoding efficiency and information preservation.
Solution Approach 2:
The patent implements feedback mechanisms where the duplicate vertex identification information and correspondence relationship information are recorded during encoding and then used during decoding to reconstruct the original non-manifold structure. This feedback loop ensures that no information is permanently lost despite the intermediate transformation.
Data Source
AI summary
An encoding method and apparatus, a decoding method and apparatus, and a device. The method of embodiments of this application includes: determining, by an encoding end, first indication information, where the first indication information is used to indicate whether geometric vertices and attribute vertices in an original mesh have a same connection relationship; determining one or two groups of non-manifold structure information based on the first indication information, where each group of non-manifold structure information includes non-manifold identification information of a target vertex and index information of a target duplicate vertex; the target vertex includes a geometric vertex and an attribute vertex in a manifold mesh; and encoding the first indication information and the one or two groups of non-manifold structure information to obtain a first bitstream.