vita
Research
DEM Error Visualization Surface Reconstruction Visualization of Social Stratification Student Research & Projects GIS & CG Links

My research is a combination of Computational Geometry and Computer Graphics/Visualization, as applied to Geographic Information Systems (GIS). The main areas currently under investigation are:

  • DEM Error - many DEMs are constructed from less-than-perfect data, and along with the computations necessary to create the final surface, errors are inevitable. This research seeks to determine the severity of the error.
  • Visualization - while many packages exist to view DEMs and other data, this research concentrates on viewing DEM error.
  • Surface Reconstruction - creating a 3D (actually 2.5D) surface from 2D sources, such as contour lines (isolines) or other sparse data arranged on a grid. The data is interpolated or otherwise used to create a digital elevation model (DEM). A DEM is useful for visualization and analysis of a particular region. Some of this work has been done in conjunction with Wm. Randolph Franklin (Electrical, Engineering, and Systems Engineering Department, Rensselaer Polytechnic Institute)
  • Interdisciplinary Research - I am in collaboration in other areas with researchers such as Mike Smith (School of Earth Sciences and Geography, Kingston University, London), John Grady (Department of Sociology, Wheaton College), and Michael Drout (Department of English, Wheaton College).

DEM and Error Visualization

A newer area of research is DEM terrain and error visualization. Traditional GIS have many capabilities to compute DEM error/uncertainty, but they require the user to go through many complex steps. Often, the result is simply a number or a 2D visualization. This research seeks to make the process much easier and faster, as well as give a better error visualization.

Our current system, DEMView (previously DEMEV; see Figure 1) has visualizations for relief, raster height colors, slope, height classes, curvature error, local elevation difference error, and local curvature difference error. The difference errors are classified according to user-defined magnitudes. The user can dynamically change the view, including the sun position, vertical exageration, and so forth. In addition, various statistics can be generated. The latest innovation is a vertical "profile cutter" that allows one to view a desired profile of two surfaces simultaneously within the context of the 3D surface visualization.

Downloads (updated 7/26/2010)

Keep in mind that this is a prototype system, used for testing algorithms and research ideas. It is not meant (yet, anyway!) to have all of the functionality of a robust commercial system. As such, it may not have all of the options available. For example, the drop-down menus are currently in a woeful state; the magnifier option has not yet been implemented.

The system was developed on Linux using an OpenGL-compatible graphics card. On such a computer, response time is quite good with grids up to approximately 1200x1200. On a system without OpenGL optimization, one can expect poor performance. Making DEMView robust across platforms is a problem we are currently working on.

The input for the system is only in ESRI ASCII Grid format.

demview1.2.gz - DEMView executable for Linux (309KB).
demview1.2.zip - DEMView executable for Windows XP (434KB).

Send email regarding execution problems and bugs to: mgousie(at)wheatonma.edu.


Figure 1

Relevant Publications/Presentations

  • Gousie, M.B., and Smith, M.J. DEMView: 3D Visualization of DEM Error. In Accuracy 2010, Proceedings of the Ninth International Symposium on Spatial Accuracy Assessment in Natural Resources and Environmental Sciences (Leicester, UK, July 2010), N.J. Tate and P.F. Fisher, Eds., ISARA, pp. 165-168. [Slides in ppt]

  • Gousie, M. B. and Milewski, S. A System for 3D Error Visualization and Assessment of Digital Elevation Models. In Proceedings of the 2007 IEEE International Geoscience and Remote Sensing Symposium (IGARSS '07) (Barcelona, 2007), pp. 4064-4067.

  • Gousie, M. B. Digital Elevation Model Error Detection and Visualization. In The 4th Workshop on Dynamic & Multi-dimensional GIS (Pontypridd, Wales, UK, 2005), C. Gold, Ed., ISPRS, pp. 42-46. [PostScript][pdf]

  • Gousie, M. B., Williams, G., Agnitti, T., and Doolittle, N. CompSurf: An Environment for Exploring Surface Reconstruction Methods on a Grid. Computers & Geosciences 29, 9 (2003), 1165-1173. Source code: CompSurf v2.0 (Developed on Linux platform). See Figure 7 below.
Surface Reconstruction Algorithms


Figure 2		 Assume we start with a digitized contour map of Mt. Washington, N.H. (Figure 2). The white space indicates areas where elevations must be computed.

A traditional method for interpolating a surface from contours is to use a partial differential equation that models a thin plate being draped over the data points (Figure 3). This yields the surface at right (colors go from blue=low elevation to red=high elevation).
Figure 3

Figure 4 		Obviously, the surface in Figure 3 is not terribly smooth. One way to improve the surface is to create a thin plate approximation instead of interpolation. This allows some of the original contour points to vary slightly, creating a smoother, albeit less accurate, surface. Furthermore, one of the reasons that the surface "scallops" in Figure 3 is because of the difference in curvature of successive contours. In our Intermediate Contour Method, we first compute new contours in between the original data. The thin plate approximation is then applied to the new data set, yielding a measurably smoother result while preserving the accuracy of the surface (Figure 4).

Following the idea that more data = better surface, we find additional data in the Gradient Lines Method. In this algorithm, "gradient paths" that follow the steepest slope from contour to contour are computed. This creates a sort of mesh that can then be filled in with various interpolating or approximating algorithms. Here, we fill in the mesh with inverse distance weighting and finish the surface with a Gaussian smoothing function (Figure 5).


Figure 5
Surface Reconstruction Animation

It takes many iterations of the thin plate method to completely fill in a regular grid of elevation points. This animation shows how a sample contour map is filled in with a thin plate approximation after intermediate contours are computed.

Relevant Publications/Presentations
  • Smith, M.J., Rose, J., and Gousie, M.B. The Cookie Cutter: A Method for Obtaining a Quantitative 3D Description of Glacial Bedforms. Geomorphology 108 (July 2009), 209-218.

  • Smith, M.J., Rose, J., and Gousie, M.B. A Method of Quantifying Subglacial Sediment Transport/Deformation. Poster presented at Geomorphology & Earth System Science, BGRG International Conference, Loughborough, UK, June 2006.

  • Gousie, M. B. and Franklin, W. R. Augmenting Grid-Based Contours to Improve Thin Plate DEM Generation. Photogrammetric Engineering & Remote Sensing 71, 1 (2005), pp. 69-79. [pdf]

  • Gousie, M. B. and Franklin, W. R. Constructing a DEM from Grid-based Data by Computing Intermediate Contours. In GIS 2003: Proceedings of the Eleventh ACM International Symposium on Advances in Geographic Information Systems (New Orleans, 2003), E. Hoel and P. Rigaux, Eds., pp. 71-77. [PostScript] [pdf]

  • Franklin, R and Gousie, M. Terrain Elevation Data Structure Operations. In 19th International Cartographic Conference & 11th General Assembly of the International Cartographic Association (ICA) (1999).

  • Gousie, M and Franklin, R. Converting Elevation Contours to a Grid. In Proceedings, Eighth International Symposium on Spatial Data Handling (1998), T. Poiker and N. Chrisman, Eds., pp. 647-656. [PostScript][pdf]

  • Gousie, M. B. Contours to Digital Elevation Models: Grid-Based Surface Reconstruction Methods. PhD thesis, Rensselaer Polytechnic Institute, 1998. [PostScript][pdf]
Visualization of Social Stratification

This current research is being done in conjunction with John Grady of Wheaton College's Department of Sociology. The idea is to use metaphors to create easy-to-use visualizations in order to see patterns in socio-economic data gathered from the US Census. The Java applets allow the user to compare different social classes, ethnic backround, job type, and income in a dynamic way. With the help of students, we have created web-based prototypes using three different metaphors:

The Target

This is the original metaphor, where the center of a dart board represents the highest income level and the rings representing progressively lower levels. Each "hit" represents 160,000 individuals at that income level.

Click to run Target 1.0

Students Sarah Milewski and Chris Stuetzle, both of the Class of 2007, implemented a newer applet of The Target which addressed some of the shortcomings of the original version.

Click to run Target 2.0

Mountain Climber

In this metaphor, the goal is to climb to the peak of a mountain, where the highest income levels exist. The different "mountains" can be moved or stacked to more easily compare desired parameters. This is a nice improvement over The Target.

This applet was implemented by Sarah Milewski '07 and Chris Stuetzle '07.

Click to run MountainClimber

Census Squared

Students Ben Burrage, Robby Grossman, Dave Machado, all from the Class of 2007, implemented this metaphor, wherein equal-size boxes are stacked on top and next to one another so as to allow easy comparisons. The more income, the more a box is filled. Although this is perhaps a less-strong metaphor, the regularity of each box makes it much easier to move about and stack to produce data patterns. This implementation also has no inherent clustering effect that both The Target and Mountain Climber have, as the area gets smaller as the income level rises in both of the latter implementations.

Click to run CensusSquared


Relevant Publications/Presentations

  • Gousie, M.B., Grady, J., Burrage, B., Grossman, R., Machado, D., Milewski, S., and Stuetzle, C. Using Metaphors in Dynamic Social Stratification Visualizations. In IV08: 12th International Conference on Information Visualization (London, 2008), IEEE, pp. 485-490. [pdf]

  • Gousie, M.B. and Grady, J. Targeting Social Stratification, presented at the International Visual Sociology Association (IVSA) Annual Conference, San Francisco, August 12, 2004. [Talk in HTML]
Undergraduate Research/Project Opportunities

There are some interesting research project or senior thesis opportunities for interested students. In particular, I need both a front and back end to my research code (described above). The front end would involve creating a nice GUI for the program. The back end would involve displaying the output graphically, allowing the user to manipulate the resulting terrain. Nate Buggia '99 developed a simple surface viewer (Figure 6), complete with lighting and shadows, among other things for a COMP 399 - Advanced Computer Graphics course.

In the summer of 2001, Trevor Agnitti '02, Nick Doolittle '03, and Greg Williams '03 tackled the problem. The resulting system is shown in Figure 7. Trevor worked on the GUI, Nick worked on surface reconstruction techniques, and Greg worked on the overall object-oriented design and much of the graphics. The system allows a researcher to view and evaluate two different surface reconstruction techniques at the same time.

The group presented their work at the Consortium for Computing in Small Colleges at Worcester State in April, 2002. The abstract "A Surface Reconstruction Research Environment" appears in the conference proceedings, The Journal of Computing in Small Colleges. This work evolved into a major research paper (source code available) in Computers & Geosciences.

Pat Sagui '04 completed a project in COMP 399 - Advanced Computer Graphics in the fall of 2003. He modeled a scene from a Lord of the Rings movie that includes many complex graphics ideas, such as lighting, textures, and fog (Figures 8 and 9).

Steven Bowe '05 did some research work for me over the summer of 2004, funded by a Mars Fellowship. He worked on finding ways to visually present errors found in digital elevation models. He presented a poster of the work at Wheaton's Academic Festival in 2005 as well as at the Consortium for Computing Sciences in Colleges at Providence College in April, 2005. His abstract appears in the proceedings as well (see below).

A nice GUI for a DEM error visualization system was implemented by Sarah Milewski '07 in the summer of 2006. This system gives traditional GIS users a quick way to check the validity of DEM data without going through many complex steps using traditional software. This ongoing research is shown in Figure 1.

Raleigh Upshur '10 worked on using a stencil buffer to zoom in on portions of data in a scene, sometimes called ``focus with context." Current research focuses on using this idea to view small places of interest on a large-scale DEM while keeping the rest of the surface in context.

Computer Graphics (COMP 365) is usually a prerequisite in order to work on these projects, but I do have others that do not require graphics programming experience.
Figure 6


Figure 7


Figure 8


Figure 9


Relevant Student Publications/Presentations
  • Upshur, R. Viewing Three-Dimensional Terrain with Focus in Context. Poster presented at the Fifteenth Annual Consortium for Computing Sciences in Colleges Northeast Conference, April 2010.

  • Stuetzle, C. Computer Modeling and Visualization of Luminescent Crystals: The Role of Energy Transfer and Upconversion. Honors Thesis, Wheaton College, 2007. See full text.

  • Bowe, S. Error Detection and Visualization in Digital Elevation Models. In Journal of Computing Sciences in Colleges, Proceedings of the Tenth Annual CCSC Northeast Conference (2005), pp. 103-104.

  • Williams, G. An Autoscheduling Optimizer for Perl. Honors Thesis, Wheaton College, 2003.

  • Williams, G., Doolittle, N., and Agnitti, T. A Surface Reconstruction Research Environment. In Journal of Computing in Small Colleges, Proceedings of the Seventh Annual CCSC Northeast Conference (2002), pp. 301-302.
Additional Presentations
  • Trying Technology on for Size. Part of the President's Commision meeting, Wheaton College, April 2010.

  • Math Connections in Computer Science. Panel discussion with Wilkens, et al. at CCSCNE, April 2009. In Journal of Computing Sciences in Colleges, Proceedings of the Fourteenth Annual CCSC Northeastern Conference (2009), pp. 57-61.

  • How Good Is Google Earth? Department of Computer Science Seminar, Siena College, Albany, NY, October 2007.

  • Can You Trust Google Earth? Department of Mathematics and Computer Science Seminar Series, Wheaton College, October 2007.

  • Math for Non-Mathers: Using Math and Programming in Everyday Life. Wheaton College Faculty Lunch Series, March 2, 2005.

  • The Role of Digital Logic in the Computer Science Curriculum. Panel discussion with Hoffman, et al. at CCSC-NE, April 23, 2004. In Journal of Computing Sciences in Colleges, Proceedings of the Ninth Annual CCSC Northeastern Conference (2004), pp. 5-9.

  • Thunderstorms, Orange Slime, and Boiling Mud, with B. Dyer and G. Collins. Part of the Puzzles in Science series, Wheaton College, 2003.

  • Implementing the Architecture, Assembly Language and Operating Systems Components of Curriculum 2001. Panel discussion with Wilkens, et al. at CCSC-NE, April 25, 2003. In Journal of Computing Sciences in Colleges, Proceedings of the Eighth Annual CCSC Northeastern Conference (2003), pp. 118-122.

  • Building a Surface Reconstruction Research Environment. Williams College Computer Science Colloquium, April 5, 2002. [Slides in PostScript][Slides in pdf]

  • Making a Mountain out of a (Math) Model. Wheaton College Faculty Lunch Series, October 26, 2000.

  • Improving Terrain Reconstruction on a Grid. Williams College Computer Science Colloquium, 1998.
GIS and CG Links (Updated 7/26/2010)

Journals/Organizations:

There are various places to get USGS data and GIS information, including:

Some Computational Geometry (CG) sources:

Last updated