- Data Acquisition
- Multiple 3D point clouds representing different views of the object are obtained from a laster scanner
- The multiple views are registered into a single set of points using an ICRP registration method
- Segmentation
- The point cloud is split into subsets representing natural surfaces of the object
- Surface Fitting
- For each point subset an appropriate surface type is determined
- A surface of this type is fitted to the point set
- At this point we only consider objects with planar, spherical cylindrical, conical and toroidal surfaces
- Model Creation
- An initial solid model is created by stitching the generated surfaces
- This initial model suffers from various naccuracies caused by
- sensing errors
- approximation and numerical errors
- possible wear of the object
- manufacturing method used to make the object
- Beautification
As our goal is to automatically reconstruct an ideal model with desired geometric regularities, those regularities have to be enforced at some stage of the process. In our approach we add a post-processing step called beautification. For this the initial model is analysed to find geometric regularities and then an improved model is reconstructed using geometric constraints.
Connector Part Example
Point Cloud and Triangulation
Segmented Mesh and Reconstructed Solid Model
We implemented some of the reverse engineering methods, but the main part of the software for creating the initial model was provided by T. Varady from the Hungarian Academy of Sciences and CADMUS Consulting and Development Ltd. Our main contributions are the concepts and software for beautification.
Cite as
Overview,http://www.langbein.org/research/solids/borg/overview/print by Frank C Langbein [26/October/2008, 16:35].
Copyright © 1995-2012, Frank C Langbein.
Except where otherwise stated, this web site, Ex Tenebris Scientia, by Frank C Langbein is licensed under a Creative Commons Attribution-Non-Commercial-Share Alike 2.0 UK: England & Wales License.