Graphics and Visualization

Our graphics and visualization research addresses the problem of converting data into compelling and revealing images that suit users' needs. Our work includes developing new representations of 3D geometry, choosing appropriate graphical realizations of data, designing software systems, and generating images for the full range of output devices.

Geometric Processing

Creating and processing geometric representations of three-dimensional objects is critical to many visual applications. In our continuing work on three-dimensional scanning of physical objects, we are developing methods for building geometric representations from measurements that include such visual attributes as color, reflectance, and self-shadowing. Our research also encompasses modeling objects in CAD systems, including creating and editing subdivision surfaces as well as using physically based methods to define geometry. Having invented fundamentally new ways to simplify and compress geometric data, we continue to devise new methods for progressively transmitting and displaying multiresolution geometries.

Interactive Data Visualization

In many domains, particularly science and business where data is often abstract, it is difficult to predict which visualization will convey the most information to a user. Often, no single view suffices, but a combination of interconnected visualizations allows the user to discover important facts. We have developed innovative approaches to visualizing complex data types that include numbers, text, and images; a wide variety of tools for increasing users' access to their data; and novel visual representations of abstract and high-dimensional data. Our research combines techniques of advanced data structures and algorithms, human-computer interfaces, and perceptual psychology to achieve new understanding of the strengths and limitations of computer visualization.

Visualization Systems

Today's computational environment requires that we design software systems that adapt continually to users' changing and growing needs. We can no longer assume that we can predict the ideal visualization for a particular domain; instead, we are providing tools that offer users the resources they need to build their own visualizations. We can no longer assume a particular hardware environment; instead, we are developing multi-user, networked systems that provide visualizations suited to the capabilities of each user and adapt to changing network conditions.

Many of today's applications are too complex for one processor or computer. For demanding visualization tasks that distill terabytes of data into graphical form, we are developing systems for geometric processing and rendering on large-scale and massively parallel systems. To enable users to exploit the availability of inexpensive computers, we are creating distributed systems for processing and rendering on clusters of workstations.

Adapting to Display Platforms

Few existing systems can exploit the latest high-resolution displays, with 200 pixels per inch. No systems exist at all to provide sophisticated visualization with the limited resources of PDAs. We are creating new tools to extend the range of visualization systems at both ends of the display spectrum.

Higher quality visualization requires more than just hardware improvements. Our research also focuses on producing higher fidelity images in all media. We are developing new algorithms for converting digital images to printed pixels and for descreening printed pictures to produce smooth digital images. We are also exploring the problems of maintaining consistent color across different media and different display devices.

Please contact Paridhi Verma to obtain copies of the Computer Science Brochure