A fully algorithmic method for geometric feature recognition (GFR), which is a frequently encountered task in developing automated software tools for concurrent engineering, is developed. A feature taxonomy based on the concept of atomic features (called PTFs) and their variations (VPTFs) enables the recognition of the infinite variety of features that arise in practice from a finite feature database. Computational economy is achieved by separating the feature extraction and feature identification phases and utilizing two new algorithms for decomposing protrusion and depression types of complex features in terms of constituent PTF/VPTFs. The algorithms are based on a heuristic understanding of the ways in which PTFs interact. The methodology results in a unique high level representation of the object in terms of context independent geometric features. The strategy has been found to be robust when applied to a large number of test objects containing a wide variety of polyhedral and cylindrical features.
No takes yet. Share an insight, caveat, or question.
Yuen et al. (1999) studied this question.