Mesh Segmentation for Sparse 3D Data Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current 3D mesh compression standards are inadequate for transmitting connectivity information and are inefficient for sparse meshes, particularly in dynamic scenes, as they require dense point clouds and are not optimized for encoding attributes of triangle faces.
Innovation Solution
A method to segment meshes into sub-meshes based on triangle properties, specifically classifying triangles by area and adjusting classifications based on neighboring triangles, merging connected components, and allowing sub-meshes to overlap by adding triangles at their borders, using artificial intelligence and machine learning to optimize thresholds per frame or sequence.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If projection-based method (V-PCC) is used to compress point clouds, then compression efficiency is improved, but the method cannot transmit connectivity information required for 3D mesh compression
Solution Approach 1:
The mesh is segmented into multiple sub-meshes based on triangle area classification. Each sub-mesh is then processed independently through the projection-based compression method. This segmentation allows the system to maintain connectivity information within each sub-mesh while still benefiting from the compression efficiency of V-PCC, as the connectivity is preserved locally within segmented regions rather than being lost in the global projection process
2Productivity
If dense point clouds are used to ensure efficient encoding, then encoding efficiency is improved, but sparse meshes cannot be efficiently encoded
Solution Approach 1:
The patent applies different processing strategies to different regions of the mesh based on local triangle area characteristics. Small triangles (detail regions) and large triangles (coarse regions) are classified and handled differently. This local quality approach allows efficient encoding of sparse meshes by adapting the compression strategy to the local density and importance of each region, rather than requiring uniform dense sampling across the entire mesh
3Adaptability or versatility
If all vertices are encoded as RAW data to handle sparse point clouds, then sparsity is accommodated, but encoding efficiency decreases and attribute information is lost
Solution Approach 1:
Instead of uniformly encoding all vertices as RAW data, the patent classifies triangles by area and applies different encoding strategies to different regions. Small triangles are encoded with higher precision while large triangles use coarser encoding. This local quality approach maintains encoding efficiency by avoiding over-encoding of less important regions while still accurately representing sparse regions where needed
4Extent of automation
If mesh segmentation is performed without prior knowledge of mesh generation, then automation is improved, but segmentation accuracy may deteriorate
Solution Approach 1:
The patent uses triangle area as a key parameter for automatic segmentation. By changing the classification threshold parameter, the system can automatically generate meaningful sub-meshes without requiring prior knowledge of the mesh generation process. The area-based classification naturally captures geometric features and boundaries, maintaining segmentation accuracy while achieving full automation through parameter-driven classification
Data Source
AI summary
Described herein is a method to segment meshes into sub-meshes based on triangle properties. The triangles are first classified according to some characteristic using their respective areas. A filtering process may change the classification according to the neighboring triangles. Then connected components are generated, and neighboring connected components are merged following a certain criteria. In some embodiments, connected components that share the most amount of edges are merged. With this technique, sub-meshes can be automatically generated without any previous knowledge of the mesh generation stage.