Alan Keahey: Visualization Portfolio

I completed my Ph.D. thesis in December '97, and am now out working in the real world. I'll try to keep these pages up to date, or transfer them to my new workspace, but they may become dated.

Most of my visualization research involves the technique of nonlinear magnification, the fundamental process central to fisheye or distortion-oriented visualizations. Below are examples (generally in reverse chronological order) of some of my visualization projects, along with some pointers to more in depth information. There is also a list of selected publications and technical reports which describe some of these items in greater detail.

Nonlinear Magnification Fields

I have developed a new model for nonlinear magnification which reduces the expression of magnification to manipulation of a scalar field of magnification values. Methods are provided which allow any nonlinear magnification system to be reduced to its scalar field representation. This new paradigm for nonlinear magnification allows application of nonlinear magnification in ways which were never before possible, such as allowing properties of the data itself to directly specify the magnification field which best magnifies those same properties. A paper with the details of these methods appeared in the Information Visualization Symposium of the IEEE Visualization '97 conference. The image below illustrates the error convergence of an iterative method for computing transformations from any arbitrary magnification field.

VRML example of a nonlinear magnification field illustrating some of the concepts described in the Vis'97 paper (VRML2.0 tested on IRIX CosmoPlayer 1.0.2, other browsers may have troubles with this).

The Nonlinear Magnification Home Page

I created and maintain the Nonlinear Magnification Home Page to serve as a central resource for information related to nonlinear magnification in all of its various incarnations. There is a brief tutorial, a list of online publications, a full bibliography, plus links to related people, projects and demos. This is the place to find out everything you wanted to know about nonlinear magnification, but didn't know where to ask.

Cluster Visualization for Data Mining

While at Los Alamos National Lab in Summer '96, I developed a system which employed nonlinear magnification for visualization of high-dimensional clusters. This system is being used by the data-mining research team of the Computer Research and Applications Group at Los Alamos to detect Medicare fraud and abuse. Here is some more information about the visualization system.

The FAD Toolkit

As part of my dissertation research, I am developing this toolkit is my primary vehicle for research on nonlinear magnification. More information about this toolkit along with execution demos can be found at the FAD Toolkit home page .

Nonlinear Image Magnification

Thanks to the modern miracle of OpenGL, it is a relatively straightforward procedure to apply the nonlinear magnification techniques I have developed to the task of image enhancement. The first example below shows how a non-occluding magnifying glass can be constructed for map visualization. This demonstrates a primary advantage of nonlinear magnification: increased resolution can be obtained while maintaining a view of the global context.

The image below demonstrates a more subtle magnification effect which can be achieved using the nonlinear magnification field techniques which I have developed. These techniques allow for very tightly controlled applications of nonlinear magnification. See a QuickTime movie (1.2 MB) showing the interactive use of the magnifying brush.

Viewing Text With Nonlinear Magnification

For a Human-Computer Interaction class, I conducted an empirical study with Julianne Marley on the effectiveness of nonlinear magnification techniques for viewing structured text files. The viewer which we used was a very simple implementation of nonlinear magnification as shown below. We were encouraged to notice that users were able to find information in structured text files faster using nonlinear magnification techniques than with other conventional text viewers. Details of this experiment are described in the Indiana University Computer Science Technical Report 459 Viewing text with non-linear magnification: An experimental study.

Hypermedia Visualization/Navigation

HyperLINK is a project I worked on at Los Alamos National Lab in Summer '95 to provide a graphical system for navigation through the WWW. A picture of the program appears below. The program was demonstrated at HyperText '96, and there is a handout with more information from that conference. In addition, there are some web pages with some information.

Database Visualization

During a research seminar on multi-media databases in Spring '95, I developed a database visualization/navigation project called Movie Space. This project represents an application of my more general research into the problem of database visualization and navigation. The tools that I am developing should prove to be useful not only for database/hypermedia navigation, but also for visualization of general graph/relational structures.

Wavelets

I am intersted in the visualization and application of wavelets. The first figure below was created from a Mathematica notebook which I wrote to visualize 2D wavelet transformations. For more details on this refer to my wavelets and image processing page.


I've also experimented with using nonlinear magnification as a means to drive wavelet transformations which create different resolution versions of images for interactive searching of sets of images. The image below shows an example of how these methods can be applied.

Selected Publications:

For more information.