Tim Stone's Professional Page

TIM STONE'S PROFESSIONAL PAGE

(This page under construction)

Recent Publications

1) "Triangulated Irregular Network (TIN) Representation Quality as a Function of Source Data Resolution and Polygon Budget Constraints", Proceedings of SPIE (The Society of Photo-Optical Instrumentation Engineers) AeroSense, Integrating Photogrammetric Techniques with Scene Analysis and Machine Vision III, SPIE, vol 3072, pp. 199-210, April, 1997.
2) "Measuring Simulated Natural Environment Terrain Fidelity", Proceedings, 1997 Distributed Simulation Symposium, pp. 1-11, September, 1997, paper 97F-SIW-022

This paper was chosen to be among the selection of (10) papers on the recommended reading list for the MIT Lincoln Labs 1997 Fall Simulation Interoperability Workshop. Furthermore, it was ranked #3 on the list.

3) "Developing a Point Selection Strategy for Elevation Data Modeling in Synthetic Environments", Proceedings, 1997 IEEE International Conference on Systems, Man, and Cybernetics, IEEE press, Vol. 3, pp. 2794-2799, Oct, 1997. ISBN # 0-7803-4053-1

4) "Automating Error Detection and Correction in Synthetic Environments", Proceedings, Simulation Interoperability Workshop, Fall 1998, paper 98F-SIW-087

This paper was chosen to be among the selection of (10) papers on the recommended reading list for the MIT Lincoln Labs 1998 Fall Simulation Interoperability Workshop.

5) "C2 (Command and Control) Experiments", IDA Document D-2194, Sept 1998

6) "Distributed Simualtion-Based Situational Awareness Experiments", IDA Document D-2409

7) "Searching for Functional Equivalence Classes in Environmental Data ",Proceedings, Simulation Interoperability Workshop, Fall 1999, paper 99F-SIW-108.

8) "Verification and Validation (V&V) of Federation Synthetic Natural Environments", Proceedings, I/ITSEC, November 2001.

This paper was chosen as the best paper at the conference in the area of Modeling and Simulation.

9) "Towards the Production of Syntactically and Semantically Correct synthetic Environment Databases", Proceedings, Simulation Interoperability Workshop, Spring 2003, paper 03S-SIW-018

This paper was chosen to be among the selection of (10) papers on the recommended reading list for the MIT Lincoln Labs 2003 Spring Simulation Interoperability Workshop.

I work in the Simulation Center at the Institute for Defense Analyses. Much of the work I do relates directly to Advanced Distributed Simulation (ADS). I am interested in the TIN (Triangular Irregular Network) creation process. This is the process by which a regular grid of points (DEM) may be approximated by drastically fewer points, but retain much of it's quality.

General Example:

Above: A shaded relief perspective of a fully utilized DEM. (Organ Mountains, Las Cruces, NM)

Below: A shaded relief perspective of a TINned DEM which uses 1/2 the points of the full DEM.

Note the degradation in the flatter areas, and the lack of degradation in the mountainous terrain.

The above example shows approximately a 50% data reduction. The original DEM had a 10 meter resolution.

What is a DEM?

A DEM (Digital Elevation Model) is an attempt to represent terrain. Consider your backyard. Imagine you want to represent your backyard digitally. One thing you could do is to go outside and measure the elevation very carefully at regular intervals, and create a grid of elevations. For instance, you might measure the elevation every 1 foot. You may end up with something like:

X X X X X X X X X X X X X X

Where each X represents an elevation you measured. This grid of points is called a DEM. You could feed it into a shaded relief rendering program and produce a shaded relief as shown above.

What is a TIN?

A TIN is a structure which is created from a DEM. Imagine now that your backyard is Texas. You thus may have millions, billions, or even trillions of elevation points you thus need to collect. If you wanted to store 1 trillion elevation points you may consume over a Terabyte of storage on your computer! Since most people do not have a Terabyte of available storage, people have been thinking of ways to minimize storage restraints. The easiest way to reduce your storage is to collect data at a sparser resolution. That is, instead of collecting a data point every 1 foot, you may want to collect one every 1 mile.

Currently, the state of the art for DEM creation is 1 meter. A 1 meter resolution DEM can take up a ton of space and consume a lot of memory to process. So, TINning becomes important. TINning is an effort to use the information stored in dense resolution grids, without the dense resolution. The basic TIN process involves:

1) Obtain a DEM.

2) Transform the DEM into a set of triangles.

The transformation may take many forms, but the general idea is that the set of triangles output from a TIN creation program contains fewer number of verticies than was in the DEM. In the above example, the number of verticies (data points) has been cut in half.

For very dense (1 meter) grids, the savings can be astounding:

An Example of DEM/TIN for 1m Terrain

Consider the following 1 meter resolution DEM:

The number of data points needed to fully represent 1 500x500 meter piece of this terrain is approximately 250,000.

Consider the following TIN of the same terrain:

This TIN uses only about 2200 points per 500x500m square.... a savings of over 99% !!!! And the quality is very very similar...These GIFs don't do it much justice, but you can see the striking similarity.

Here is the actual triangulation:

Note how the algorithm uses small triangles to describe terrain that is rugged and larger triangles to describe flatter terrain.

Back to my main page

This page hosted by
Get your own Free Home Page