PhD. Thesis of Manolya E. McCormick:
Interactive Freeform Editing Techniques for Large-Scale, Multiresolution Level Set Models

Abstract:
Level set methods provide a volumetric implicit surface representation with automatic smooth blending properties and no self-intersections. They can handle arbitrary topology changes easily, and the volumetric implicit representation does not require the surface to be re-adjusted after extreme deformations. Even though they have found some use in movie productions and some medical applications, level set models are not highly utilized in either special effects industry or medical science. The lack of interactive modeling tools makes working with level set models difficult for people in these application areas.

This dissertation describes techniques and algorithms for interactive freeform editing of large-scale, multiresolution level set models. Algorithms are developed to map intuitive user interactions into level set speed functions producing specific, desired surface movements. Data structures for efficient representation of very high resolution volume datasets and associated algorithms for rapid access and processing of the information within the data structures are explained. A hierarchical, multiresolution representation of level set models that allows for rapid decomposition and reconstruction of the complete full-resolution model is created for an editing framework that allows level-of-detail editing. We have developed a framework that identifies surface details prior to editing and introduces them back afterwards. Combining these two features provides a detail-preserving level set editing capability that may be used for multi-resolution modeling and texture transfer. Given the complex data structures that are required to represent large-scale, multiresolution level set models and the compute-intensive numerical methods to evaluate them, optimization techniques and algorithms have been developed to evaluate and display the dynamic isosurface embedded in the volumetric data.


Last modified on January 10, 2013.