Frank C Langbein
Ex Tenebris Scientia
Connector Example

Reverse engineering a physical object is the extraction of information from a particular object which is sufficient to reproduce it. In most cases only an approximation of some aspect of the original part is required. For this research project we are interested in reverse engineering the shape of mechanical parts which can be described by planar, spherical, cylindrical, conical and toroidal surfaces. Our goal is to create a system that can automatically generate a complete and consistent solid model for a given, simple mechanical part which has all the important geometric properties of its real instance.

It is expected that a system like this has various practical applications by dramatically reducing the amount of human interaction required for generating CAD models. Often CAD models of existing parts do not exist or are not available in a usable form or a CAD model of a non-computer generated model like a clay model is required. Applications for those models range from design and re-design to manufacturing to analysis. An important aspect for many of these applications is that an accurate model with the intended shape properties is generated.

For a detailed describtion of the beautification system and the algorithms, etc. please see Frank C. Langbein's PhD thesis Beautification of Reverse Engineered Geometric Models.
Overview of the Reverse Engineering System
To obtain the initial data for a part we use a commercial 3D laser scanner. From the generated point clouds a valid CAD model can be created by using existing technologies. This can be done by segmentation of the point sets into subsets and by finding surfaces that best approximate those subsets. From these surfaces a valid boundary representation model can be created.
Our Strategy for Beautification
To ensure that the model generated by the system exhibits certain geometric regularities we add a post-processing step, which we call beautification. In this step the initially created boundary representation model is analysed for regularities and a suitable subset of regularities is selected to generate an improved model using geometric constraints.
Publications and Presentations
More details about our methods are available in various publications and presentations.
The software developed by us for the beautification step and some parts of the other reverse engineering phases is available under the GNU General Public license. We also provide a set of models used to test our system.



This project is supported by the UK EPSRC Grant GR/M78267. We would like to thank T. Varady from the Hungarian Academy of Sciences and CADMUS Consulting and Development Ltd. for providing reverse engineering software.

Cite as Beautification, [26/October/2008, 17:19].
[Contact Details]