Author | Ottmar Klaas, Mark Beall, Mark S. Shephard |
---|---|
Title | GENERATION OF GEOMETRIC MODELS AND MESHES FROM SEGMENTED IMAGE DATA |
Year | 2014 |
Journal | NAFEMS Americas Conference 2014 |
Abstract | Domain descriptions for numerical analyses originating from 3D images pose significant challenges as they have to be converted to a discretization suitable for a numerical analysis. Methods that perform this process by converting the image data directly to an analysis mesh suffer from a number of disadvantages with respect to both the flexibility to create well controlled meshes, and with respect to the ability to effectively represent the heterogeneous material regions of the imaged domain. We will present an alternative approach that avoids disadvantages through the construction of a non-manifold model representation. This paper presents a set of automatic procedures for the construction of a non-manifold model and mesh given a segmented image data set. The process of creating a non-manifold model first allows the user to utilize the same meshing procedures that are applied to CAD models. The procedures developed are split between (i) operations on the voxel data set and (ii) operations on the non-manifold model. The former deal with idiosyncrasies of the voxel data which are getting in the way of generating a suitable non-manifold model. The procedures developed for the non-manifold model eliminate voxel artifacts by smoothing the surfaces with a specifically designed shape smoothing procedure. |
PDF File | Download |