In January 2007, I successfully defended my Masters Thesis entitled Synthesizing techniques based on multiresolution. And in July 2007, I received my Masters of Science from the University of Calgary. I very much enjoyed my time at the Jungle where I studied under the supervision of Dr. Faramarz Samavati. Not only did teach me a lot about Science, Math, Graphics, and Research ... he also taught me about life. For everything I learned from him, I am in-debted.
Contextual void Patching for Digital Elevation Model, Wecker, L., Samavati, F.F. and Gavrilova, M., The Visual Computer, accepted, in press, 2007.
Iris Synthesis: A MultiResolution Approach, Lakin Wecker, Faramarz F. Samavati and Marina Gavrilova, GRAPHITE 2005 Conference, short paper, New Zealand, December 2005.
Digital models can be created by gathering a set of measurements from geometric objects. For various reasons, these models may be incomplete representations of the objects. Incomplete models fail to meet the requirements defined by their potential applications.
In this thesis we develop a multiresolution approach to synthesizing digital models. Multiresolution can reduce the number of data points in a model to produce a coarse approximation. The coarse approximation will lack some features that are present in the original data. These features can be captured and are usually called the details. We consider the details to be characteristics of the model and we consider the coarse approximation to be the overall structure of the model. Using MR, we can iteratively decompose the model to form a hierarchy of components. We selectively combine components from existing models to synthesize new models with the appropriate characteristics.
Our technique can synthesize models with characteristics similar to thosefound in existing models. The technique is general and can be adapted for many application areas. As a demonstration we provide two sample applications. First, it is used to synthesize patches for the voids commonly found in digital elevation models (DEM). Secondly, it is used to augment existing iris image databases by synthesizing iris images. The results in both applications demonstrate that our approach is effective at synthesizing realistic models.