Sketch-based Semantic Feature Modeling With Q-Complex Data Structure.

摘要

Considering design efficiency, it is desirable to construct a 3D product model with user-friendly sketch inputs and by reusing existing knowledge.;For definition, a semantic feature is defined as a group of related abstract entities, relations, and rules, to support sketch. The abstract entities are augmented with sketch entities to support sketch-based feature retrieval; relations are extended with inclusion relation and corresponding relation to support sketch-based feature instantiation; rules are added to encode domain knowledge to resolve a sketch's ambiguity. A semantic feature model is constructed from conceptual layer, to feature layer, and to sketch layer. Differently, its feature layer contains a set of feature subspaces and its shape is detailed via sketch by corresponding relations.;For representation, a new non-manifold boundary representation (B-reps)---Q-Complex with inclusion topology, is proposed to represent semantic feature models. Inclusion topology, allowing one entity to be embedded into another entity, is used to explicitly represent the relative positions among semantic features. With Q-Complex, its efficiency of 36 basic queries is several times faster than most-state-of-the-art B-reps, while the storage cost is compact in some situations.;For creation, a semantic feature language containing vocabulary, sketch-based operators, and grammars is design to express the modeling process as an algebraic expression. To reuse predefined semantic feature, a hybrid retrieval algorithm is proposed based on semantics and sketches. The semantic feature library is subdivided into feature subspace and each is label with a semantic synonym-set, to support semantic retrieval. Also, each feature is encoded with n-connecting scheme to support stroke-by-stroke sketch retrieval.;A proof-of-concept system---FeatureSketch, is proposed and applied to jewelry design and mechanical design, with examples to demonstrate its usage.