|Title:||An Evolutionary Strategy for Free Form Feature Identification in 3D CAD Models|
|Authors:||Langerak, T. R.|
Vergeest, J. S. M.
|Citation:||WSCG '2007: Full Papers Proceedings: The 15th International Conference in Central Europe on Computer Graphics, Visualization and Computer Vision 2007 in co-operation with EUROGRAPHICS: University of West Bohemia Plzen Czech Republic, January 29 – February 1, 2007, p. 145-152.|
|Publisher:||Václav Skala - UNION Agency|
|Document type:||konferenční příspěvek|
|Keywords:||rozpoznávání znaků;evoluční algoritmy;znaková knihovna|
|Keywords in different language:||feature recognition;evolutionary algorithms;feature library|
|Abstract:||Nowadays, most 3D CAD systems support the use of form features. The main advantage of form features is that they provide parametric, high-level support for shape manipulation. When the parametric information of a shape is not available, it can be retrieved using a feature recognition procedure. In this procedure, a target shape is recognized as an instance of one or more features in a pre-defined feature library. The speed and accuracy of the feature recognition procedure significantly improve when the target feature type is known. This information can be provided by a user, but we propose a new method that identifies the feature type automatically. This method uses an evolutionary algorithm to find the optimal feature type. The algorithm randomly generates a population of instances of every type of feature available in the feature library. From this initial population of feature instances, successive populations are generated using the principle of natural selection. The algorithm was tested for different settings and was found to correctly identify features in between one and three minutes.|
|Rights:||© Václav Skala - UNION Agency|
|Appears in Collections:||WSCG '2007: Full Papers Proceedings|
Files in This Item:
|Langerak.pdf||Plný text||430,39 kB||Adobe PDF||View/Open|
|Langerak_prezentace.ppt||Prezentace||663,5 kB||Microsoft Powerpoint||View/Open|
Please use this identifier to cite or link to this item:
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.