Feature Recognition in B-Rep Models via Face Adjacency Graphs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing feature recognition techniques in solid modeling, particularly graph-based approaches, are computationally expensive and inefficient when dealing with feature interactions, and often require specific feature hints that may not be generally applicable, especially in cases where feature information is missing or not directly available in design trees.
Innovation Solution
A computer-implemented system that identifies topological features in Boundary Representation (B-rep) models by selecting faces, determining boundary edges, and generating output highlighting these features, which includes algorithms for recognizing revolved features, pockets, islands, slots, and feature chains, enabling automated extraction of feature information from three-dimensional models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If graph-based approaches are used for feature recognition, then feature information can be extracted from B-rep models, but the computational cost increases significantly especially in the presence of feature interactions
Solution Approach 1:
The patent segments the feature recognition process into two distinct phases: (1) generating a face-adjacency graph representing the entire B-rep model, and (2) matching feature templates against subgraphs of this pre-built structure. This segmentation allows the computationally intensive graph construction to be performed once, enabling multiple feature queries to be executed efficiently without repeatedly processing the entire model geometry.
Solution Approach 2:
The patent performs preliminary action by pre-processing the B-rep model to construct a complete face-adjacency graph before feature recognition begins. This pre-built graph structure includes all topological relationships between faces, edges, and vertices, allowing subsequent feature matching operations to proceed by simple subgraph searches rather than complex geometric analyses, significantly reducing real-time computational requirements.
2Adaptability or versatility
If traditional feature recognition methods are used, then features can be identified, but the methods require specific feature hints that are not generally applicable and may not be available in design trees or during CAD translation
Solution Approach 1:
The patent implements a universal face-adjacency graph framework that can represent any topological feature type (pockets, slots, protrusions, holes, etc.) without requiring feature-specific preprocessing or hints. The graph structure universally captures all face-to-face adjacencies and geometric attributes, allowing the same matching algorithm to recognize diverse feature types by comparing against different template graphs, thereby eliminating the need for feature-hint dependencies.
Solution Approach 2:
Instead of starting with feature definitions and trying to match them to model elements (which requires feature hints), the patent inverts the approach by building a complete adjacency graph of all model elements first, then searching for feature patterns within this graph structure. This inversion allows the system to discover features directly from topological relationships without needing prior knowledge of what features to look for or how they are represented in the design history.
Data Source
AI summary
A computer-implemented product including instructions embodied in a non-transitory computer read-able medium that, when executed by a processor, cause the processor to receive input selecting one of a plurality of faces of a boundary representation model of an object. The computer-implemented product may include further instructions to identify boundary edges of the model outside the one of the faces not separated from the one of the faces by an intervening concave or convex edge. The boundary edges may define a perimeter of a topological feature containing the one of the faces. The computer-implemented product may include further instructions to identify a set of the faces including the one of the faces contained by the perimeter and generate output highlighting the boundary edges and set of the faces of the topological feature.