In this paper the problem of matching freeform surface features against measured point data is studied. The motivation for this investigation is the recent demand for fast and robust algorithms to detect, to recognize and to automatically parameterize predefined types of shape in existing objects. These algorithms serve for the quick and accurate reuse of existing shapes in new designs of products in computer-aided design (CAD). Starting from a theory of parameterized shapes, or features, we developed a practical method and algorithm to determine the occurrence of a freeform feature type in measured data, and to obtain the parameter values of the instance. The parameters not only control the optimal placement and scale of the instance but also some characteristics of the shape itself. We have investigated the influence of 1) the density of the digitized points, 2) the representation form of the shape feature, 3) the matching criterion, 4) the optimization strategy. In addition, perfor...
Joris S. M. Vergeest, Sander Spanjaard, Jos J. O.