HPC2N
High Performance Computing Center North
Theme: Scientific Visualization
High Performance Computing Center North (HPC2N) invites to seminars in Scientific Visualization at Umeå University. The seminars will be given by Professor Roger Crawfis, Department of Computer and Information Science, The Ohio State University, USA. (http://www.cse.ohio-state.edu/~crawfis/) |
13.15 - 14 | Seminar 1: Volume Models for Engineering Design and Analysis |
14.15 - 15 | Seminar 2: Interactive Visualization of Large Datasets |
15.00 - 15.30 | Coffee break (served outside MA121) |
15.30 - 16.15 | Seminar 3: Virtual surgerical simulators |
Abstract
In this talk, I will describe research into converting surface models, such as a CAD model, into a volumetric form. This process is known as voxelization. CAD models provide accurate specification for an underlying surface model. These surface models are often complex, having surface specifications comprised of analytical basis functions that are difficult to deal with. A more problematic surface model discription is that of Contructive Solid Geometry (CSG) models. These models are specified as using analytical functions, as well as analytical operators that combine or subtract these functions to generate a surface model. Frequently, these models have to be converted to a more simple model, composed of geometric simplices. For sufficient accuracy, this conversion often generates extremely large triangle approximations. Voxelizing these large triangular meshes is computationally expensive. Volume models are useful, in that they cast the 2D surface model into a three-dimensional implicit model. With this 3D model, we can query properties of the space around the surface. Many contact simulation, die-casting problems and material analysis problems can best be studied in this volume framework. This talk will describe research to voxelized such models and a distance volume model that allows for the extraction of equi-distant surfaces from the original surface with a guaranteed error-bound.
Abstract
In this talk, I will describe research into converting surface models, such as a CAD model, into a volumetric form. This process is known as voxelization. CAD models provide accurate specification for an underlying surface model. These surface models are often complex, having surface specifications comprised of analytical basis functions that are difficult to deal with. A more problematic surface model discription is that of Contructive Solid Geometry (CSG) models. These models are specified as using analytical functions, as well as analytical operators that combine or subtract these functions to generate a surface model. Frequently, these models have to be converted to a more simple model, composed of geometric simplices. For sufficient accuracy, this conversion often generates extremely large triangle approximations. Voxelizing these large triangular meshes is computationally expensive. Volume models are useful, in that they cast the 2D surface model into a three-dimensional implicit model. With this 3D model, we can query properties of the space around the surface. Many contact simulation, die-casting problems and material analysis problems can best be studied in this volume framework. This talk will describe research to voxelized such models and a distance volume model that allows for the extraction of equi-distant surfaces from the original surface with a guaranteed error-bound.
Abstract
A critical component of air flight training is now the requirement of
many hours of simulated flying on a professional flight simulator.
These simulators have greatly reduced the time and cost of in-flight
training and improved the safety level of current pilots. Without the
use of such simulators, much more expensive equipment and personnel
time would need to be expended. As our ability to manage and display
large three-dimensional datasets increases, the medical community is
actively pursuing a similar shift or improvement in training. This
talk with discuss the critical factors in developing surgical
simulators and other training tools for the medical community. A
common thread thru all of these is the need to handle volumetric data,
real-time frame-rates and more attention to realistic rendering. These
systems also require interactive sounds and haptic feedback.