(Dr. Thomas R. Kurfess, advisor)
"Surface Extraction from Coordinate Measurement Data"
Advancing technologies for the mechanical acquisition of coordinate measurement data have made it possible to gather dense quantities of copious data that more completely describe a measured part. The data supplied in this manner typically require preparation and manipulation before they can be used directly by the analysis procedures. This adds an intermediary step to the analysis.
This thesis proposes a method of surface extraction to provide both automation of the preprocessing step and enhanced measurement capabilities. Methods for the triangulation of the data, recognition and extraction of unique surfaces, and correlation of those surfaces to a CAD model are developed and discussed. A software application is developed and implements these methods, demonstrating their capabilities.
The results of this implementation indicate that surface extraction from coordinate measurement data enhances measurement capabilities by creating point-to-surface associations and by approximating those surface definitions, both without need for a CAD model. Further, by correlating the surfaces with an available CAD model, established inspection techniques are supported by providing an initial alignment and identifying possible sets of irrelevant data to be omitted from the analysis. Improvements to the reliability and accuracy of the pattern recognition and surface definition schemes are still necessary, as well as improvements to the efficiency of the correlation procedure.