Scott E. Dillard

Institute for Data Analysis and Visualization
University of California, Davis
sedillard at ucdavis dot edu

Research______________________________________

Construction of Simplified Boundary Surfaces from Serial-sectioned Metal Micrographs
Scott E. Dillard, John F. Bingert, Dan Thoma, Bernd Hamann
IEEE Visualization 2007, Sackatomatos, California

This project, supported by the Materials Design Instute of Los Alamos Nat'l Laboratory, introduces a method for constructing boundary surfaces of cell-like objects from an input stack of planar boundary curves. We use a simple physical model to interpolate the boundaries between the slices, then extract a triangle mesh with a combined smooth-and-simplify algorithm which removes aliasing artifacts and reduces the surface complexity by an order of magnitude.
PDF Slides

Tessellation of Quadratic Elements
Scott E. Dillard, Vijay Natarajan, Gunther H. Weber, Valerio Pascucci, Bernd Hamann
17th International Symposium, ISAAC 2006, Kolkata, India, December 18-20, 2006. Proceedings, pages 722-731.

In this paper we prove a simple tessellation scheme for triangular piecewise-quadratic elements which breaks them into monotone peices, yielding a mesh with a level-set topology equivalent to the same mesh with piecewise-linear elements. This allows the construction of Reeb graphs, contour trees or other topological structures from piecewise-quadratic triangle meshes.
SpringerLink Conference version Extended tech report

Topology-Controlled Volume Rendering
Gunther H. Weber, Scott E. Dillard, Hamish Carr, Valerio Pascucci, and Bernd Hamann
IEEE Transactions on Visualization and Computer Graphics. 13 (2), pp. 330-341. 10.1109/TVCG.2007.47

This paper describes how the contour tree can be used to define transfer functions for volume rendering 3D scalar fields, so that topologically distinct features in the data can be given different optical properties. We give a hardware-accelerated volume rendering algorithm for datasets given on 3D rectilinear grids with trilinear interpolation. This interactive technique allows the contour tree to be used as an interface element for exploration of complex 3D datasets. (Note that the contour tree link goes to Hamish Carr's webpage, not mine.)
IEEExplore UC pdf
Here is a movie illustrating the topological simplification method described in the paper. avi mov

Software_____________________________________

tltree
Compute the contour tree of a 3D image with trilinear interpolation. This is a 32-bit Linux binary (GNU libc 6). I might get around to making a Windows binary (I don't have a Mac, sorry) but the source for this program is included in the libtourtre library, below. This program is fast and robust, but currently uses a lot of memory. To give you an idea, it processes the 256^3 Bonsai dataset at volvis.org in 50 seconds using about 430 megs of ram on a 2.16ghz 32 bit laptop.

libtourtre
This is a library I wrote to collect the contour tree code I had lying around after working on the above mentioned projects. It computes the contour tree from a function that you (the app writer) provide. It's in simple C, and comes with an example application. I've used it in at least one project, but I haven't beaten all the bugs out of it yet. For one, the error messages may be as terse as "assert(false)". Feel free to email me with questions.

Vec
This is a Haskell library for small arrays of numbers, like vectors or points, and some basic linear algebra operations for them. Darcs repository is here.