3D Model Comparison Using Hierarchical Surface Descriptors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing 3D model comparison and search systems face challenges in efficiently cataloging, comparing, and displaying similarities and differences between 3D models, particularly when dealing with large datasets and varying geometric representations, and lack the ability to prioritize part-based searches.
Innovation Solution
A method that constructs descriptors capturing complete 3D model and face-level properties, allowing for precise comparison and display of similarities and differences across multiple reference frames, using geometric and topological features, and a reconciliation process to match and qualify parts of 3D models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If traditional 3D model comparison systems use a single reference frame or best fit approach, then the comparison process is simplified, but the ability to identify differences across multiple coordinate systems is lost
Solution Approach 1:
The patent segments the comparison process into multiple independent reference frames instead of using a single reference frame. Each reference frame maintains its own coordinate system, allowing comparisons to be performed locally within each frame while preserving the ability to identify differences accurately across different coordinate systems.
Solution Approach 2:
The patent adds a temporal or multi-frame dimension to the comparison process by introducing multiple reference frames simultaneously. This allows the system to compare 3D models across different coordinate systems without collapsing them into a single best-fit reference frame, thereby maintaining measurement precision while managing complexity.
2Measurement precision
If 3D model search systems use comprehensive descriptors for all model properties, then search accuracy is improved, but computational time and processing resources increase significantly
Solution Approach 1:
The patent segments descriptors into hierarchical levels (complete model descriptors, part descriptors, and face descriptors). This segmentation allows the search system to first filter models using coarse-grained complete model descriptors, then apply finer-grained part and face descriptors only to candidate matches, thereby maintaining search accuracy while reducing overall computational time.
Solution Approach 2:
The patent applies partial action by using different levels of descriptor detail selectively. Instead of applying comprehensive descriptors to all models in the database, the system uses simplified descriptors for initial filtering and applies detailed descriptors only when necessary, reducing processing time while maintaining search accuracy for final results.
3Measurement precision
If the system processes complete 3D models with high detail, then comparison precision is improved, but the ability to handle large datasets efficiently is reduced
Solution Approach 1:
The patent segments 3D models into hierarchical components (complete models, parts, and faces). This segmentation enables the system to process and compare models at different levels of detail appropriately. Complete model descriptors provide overall comparison precision, while part and face descriptors enable efficient filtering and detailed comparison only where necessary, improving both precision and productivity.
Solution Approach 2:
The patent applies partial processing by using high-detail descriptors for complete models only when necessary for final comparison, while using simplified descriptors for initial database searches and filtering. This approach maintains comparison precision for critical operations while improving overall dataset processing efficiency by avoiding unnecessary high-detail processing throughout the entire workflow.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention discloses a method for indexing 3D digital models, retrieving same, comparing same and displaying the results thereof in a 3D space. The method includes four complementary steps, i.e. display, comparison/search, surface matching and result classification. Said steps can be embedded within one another or can be implemented separately. The implementation of a process for searching/comparing 3D models based on descriptors partially relating to the surfaces requires a process of surface matching (pairing surfaces between the 3D models).