Automatic reconstruction of CAD models and properties from digital scans

Fausto Bernardini, Purdue University

Abstract

Creating geometric models of objects from digital scans has a vast range of applications, such as reverse engineering, prosthesis design and creation of 3D content for virtual environments. In this thesis, we review prior work in the field and analyze the limitations of current state-of-the-art methodologies. We then describe a novel set of techniques, based on alpha-shapes and algebraic patch fitting, to reconstruct models of solids from unorganized sets of points. Our methods rely only on the knowledge of the coordinates of a sufficiently dense point-sampling of the object's surface. We give a formal definition of the problem and sufficient conditions for the sampling to allow a consistent and accurate reconstruction. We also discuss practical issues and describe methods to handle less-than-ideal scans. Our algebraic patch fitting schemes allow the reconstruction of smooth surfaces as well as objects with sharp features and adapt their level of detail to the local complexity of the model. We also describe octree-based hierarchical spline modeling of dense, volumetric data for visualization and interrogation of large data sets. The implementation and application of these techniques to the modeling and visualization of geometry and physical properties in virtual environments are illustrated. We conclude by discussing open problems and directions for future work.

Degree

Ph.D.

Advisors

Bajaj, Purdue University.

Subject Area

Computer science

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS