|M.Sc Student||Ben-Shabat Yizhak|
|Subject||Adaptive Hierarchical Multi Resolution Volumetric|
Geometric Modeling of Porous Micro-Structures
|Department||Department of Mechanical Engineering||Supervisor||Professor Anath Fischer|
|Full Thesis text|
Three dimensional reconstruction and modeling of micro structures from scanned data had become an area of research in many disciplines in general and in particular in bio-mechanical micro-structures such as bone structures. One of the main challenges when working with the volumetric representation of a micro-structure is the large amount of data required to accommodate for the high level of detail. Large amounts of data require long processing time and massive computer resources. In some cases, an approximated low resolution model, which contains significantly less data, is adequate. Constructing a multi-resolution model overcomes this problem in the cost of memory space. The disadvantage of working with such approximated models is the loss of information and accuracy. Yet, when using multi-resolution models it is possible to control these disadvantages adaptively and optimally.
A multi-resolution model stores the original high resolution model with several gradually descending low resolution models in a hierarchical structure. This enables a system or a user to choose the level of detail which is required by the application, for example according to porosity criterion which characterizes micro-structure. Furthermore, it is possible to construct an adaptive multi-resolution model by combining information from different levels of detail. Constructing an adaptive model is possible by defining a local region of interest (ROI) and optimally defining its level of detail according to geometrical criteria such as curvature. When constructing a multi-resolution model in a straightforward approach, its topology changes significantly. This challenge is addressed by utilizing a topological preservation scheme which is integrated into the hierarchical data structure. Topological preservation enables lower resolution models to preserve the topological properties and the shape characteristic of the original model.
The main contribution of this research is a new CAD method for constructing adaptive multi-resolution models while preserving the topology in general, and more specifically in bone porous micro-structures. The feasibility of the method is demonstrated on several examples.