![[MAPSLab Papers]](/image/papers.gif){kind=small}
in: AUTOCARTO 11: International Symposium on Computer Assisted Cartography, Minneapolis, Minnesota, October 30-November 1 1993, pages 267-277.
|
Large scale distributed simulation has been used to support training
for battalion level operations within the DoD/Army's SIMNET
facility. One of the key components for realistic combat training,
through the use of hundreds of low-cost, high realism simulators, all
linked using a variety of telecommunications technologies, is the
fidelity of the underlying shared environment. For ground-based
simulations the environment includes the terrain, road networks,
buildings, and natural features such as drainage, forests, and surface
vegetation. Given the limitations in graphics rendering capabilities
for the low-cost computer image generation systems associated with
SIMNET, one must trade fidelity in terrain representation for
feasibility of simulation.
Our work has focused on reducing the number of polygons that must be
rendered in a real time simulation while improving the overall
fidelity of the terrain visualization. In this paper we present a new
method for the generation of a terrain representation beginning with a
digital elevation model (DEM) and producing a triangular irregular
network (TIN) suitable for scene generation. We compare our technique
to two commonly used methods, VIP and LATTICETIN and
provide a performance analysis in terms of residual error in the TIN,
and in the number of points required by each method to achieve the
same level of fidelity. We also briefly outline the use of selective
fidelity to improve the level of detail in selected sub-areas of the
DEM.
While motivated by the constraints of real-time distributed simulation, this research has applications in many areas where an irregular network is a more computationally efficient representation than a dense digital elevation model.
|
![[Home]](/image/home.gif){kind=small}
maps+webmaster@cs.cmu.edu
(last updated 14 Sep 1998)