Digitized data segmentation and CAD modeling for reverse engineering.

摘要

In many cases, a product design starts with a physical model. The CAD model is then extracted from the physical model. Moreover, many products have been manufactured in the past and do not have an associated CAD model. To redesign or manufacture such a product, a CAD model must be available. The creation of a CAD model for a prototype or an existing product is called reverse engineering. Reverse engineering, for the most part, is performed as an interactive process where the designer identifies the surfaces features from digitized data and models the surfaces accordingly.; In this research, an automatic approach for defining part features from scanned data has been developed. A new segmentation approach that employs neural networks is proposed for extracting primitive surfaces suitable for surfaces fitting. The segmentation approach integrates edge and region based approaches which eliminates the noise produced by incorrect neural network classifications. Moreover, a procedure for fitting quadratic and free form surfaces using neural networks is proposed. The proposed segmentation and surface fitting approaches have been verified and compared with existing approaches using real and computer generated scanned data.