Shape feature definition and recognition in Internet-based product design

In the current research, a shape feature definition language (FDL) and corresponding feature recognition (FR) algorithm are presented to support product design in Internet-based CAD (ICAD) systems.; A shape feature is a combination of topological and geometrical entities on the solid model of a part that performs a certain function. Examples of features are rib, slot, hole, etc. CAD models do not explicitly contain shape feature information. Therefore it has to be extracted via FR.; In the Internet-based CAD system where the FR becomes an application service provider to users, it must be able to recognize shape features for a variety of users with different requirements with respect to shape features. For example, features required by machining (such as a slot) are different from features required by injection molding (such as a rib).; The FDL presented in this research is designed to solve the above problem. In addition to define a feature using geometry, topology and attributes, like existing approaches, the feature spatial relation can also be defined (such as “ON”, “IN” and “ADJACENT”). Therefore, design rules about how features interact with each other can now be checked, which cannot be done in previous research. The FDL is written in XML that can be easily browsed, edited and exchanged over the Internet.; A feature decomposition network (FDN) approach is presented to recognize feature in a more efficient way than conventional methods by reducing redundant subgraph-matching operations in graph-based feature recognition.; The FDL also provides a useful mechanism for archiving and retrieving CAD models from a design database at feature level. Maximum common subgraph (MCS) is used to evaluate the similarity between two models. An approximate MCS search algorithm is presented to search the MCS more efficiently than conventional approaches.; A prototypical ICAD system is implemented. The usefulness the FDL and the performance of the feature recognition and feature-based CAD model retrieval approaches are tested using this prototype system.