Interrante/Chen graphics groups: (l-r) Minh X. Nguyen, Seonho Kim, Xiaoyan Cheng, Gary T Dahl, Gabriele Gorla, Haleh Hagh-Shenas, Timothy M. Urness, Sunghee Kim, Alex Liberman, Victoria Interrante, Feng Li. Not pictured: Hae Young Kim, Che-Hung Kuo, and Baoquan Chen.

The need for effective visualization of data in science and medicine, the heightened expectations of users for high quality graphics, and the growing importance of multimedia give some indication of the importance of the field of graphics. The Department of Computer Science and Engineering is fortunate to have recently hired two active young researchers in this area, Victoria Interrante and Baoquan Chen. They each have many exciting projects currently underway, many involving graduate students, thus forming a growing graphics group in the department.

Graphics project 1c: Mapping a 2D texture onto a 3D surface without seams or distortions. Researchers: Gorla, Interrante, and Sapiro.

Assistant Professor Victoria Interrante, who joined the department in 1998 and is the recipient of the prestigious Presidential Early Career Award for Scientists and Engineers, brings her knowledge of art and interest in visual cognition to the science of graphics to focus on effectively communicating information through images. She is interested in shape and depth perception, shape-based feature extraction, and the representation of 3D shape through texture.

Several research projects are directed towards conveying 3D shape via lines of principal curvature. The aim is to improve visualization of shapes using textures wrapped around the shape. A project with sabbatical scholar Jack Goldfeather, Carleton College, addresses the problem of computing accurate and smoothly varying estimates of principal directions at points on the surface. Another problem in this area is to determine the kinds of texture that can aid an observer in accurately perceiving a shape. Graduate student Sunny Kim is assisting in this study. A third line of research, pursued with graduate student Gabriele Gorla and Associate Professor Guillermo Sapiro (Electrical and Computer Engineering), aims to map a texture pattern onto an arbitrary 3D shape so that the result is seamless and free of distortions, and so that the dominant direction of the texture is aligned with the direction of strongest surface curvature.

A visualization project involving multivariate datasets focuses on the problem of showing multiple distinct values simultaneously at each point in space. For example, on a map one might want to show at each point a number of different kinds of data to allow the viewer to see any relationships among the data. Color can be used to effectively represent at most two or three values at each point. Interrante, assisted by undergraduate Hae Young Kim, is investigating the use of texture variations to add more dimensions to the multivariate visualization. (See figure at page bottom.)

Numerical simulations of turbulent flow and other complex phenomena produce terabytes of data, far too much to examine. Currently much is just discarded. Interrante, collaborating with Professors Ivan Marusic and Graham Candler in Aerospace Engineering and Mechanics, and with assistance from graduate students Jason Gott and Seonho Kim, is working on methods to automatically identify key features in the mass of data, such as the evolution of individual vortices in a turbulent boundary layer.

Graphics project 3: Extracting individual vortical structures in a turbulent boundary layer for the quantitative analysis of turbulent flows. Researchers: Interrante, Marusic, Candler, Gott, and S. H. Kim.

Interrante is working with Professor Al Yonas of the Institute for Child Development and Dr. Stephen Christiansen, Director of Pediatric Ophthalmology, on a project whose goal is to develop automatic techniques for screening photographs of infants and young children for indications of strabismus. Strabismus is a condition in which the eyes fail to work in a coordinated fashion. Without early treatment, it can lead to a permanent loss of stereo vision ability, or even to a permanent loss of vision in the nondominant eye. The student working on this is Sunny Kim.

Graphics project 1a: Investigating the effect of texture orientation on perception of 3D shape. Researchers: Interrante and Kim.

Assistant Professor Baoquan Chen, who joined the department fall semester, is interested in more efficient high-quality rendering of complex scenes, especially those composed of both scenes modeled by polygons and those modeled with discrete data such as clouds, smoke, and 3D medical images (e.g., MRI, CT). The two approaches to improving rendering speed are better algorithms and better hardware. Chen has worked on both approaches in the past and continues to do so. One of his current projects involves the design of an architecture that will render a mixture of polygon and volumetric data.

Graphics project 2: Representing multi-variate data with texture pattern variations. Researchers: Interrante.

On the algorithm side of the problem, one approach, view dependent adaptive isosurface extraction from volumes, tries to extract only polygons that most effectively contribute to the view desired, for example, brain tissue in an MRI, thus reducing the number of polygons that have to be processed. Graduate student Xiaoyan Cheng is assisting in this project. The choice of representation of the scene, whether polygon or discrete point, can affect the efficiency of processing. In general, a combination of polygon and discrete representations can be most appropriate. Graduate students Minh Xuan Nguyen and Haleh Hagh-Shenas are assisting in this work which aims to determine the most effective representation of a scene for a desired view.

A well-known problem in graphics is the aliasing problem which can occur when 3D primitives (polygons or discrete points/pixels) are projected to the screen and the data sampling skips important information resulting in an image which is not correct. Currently anti-aliasing algorithms use a uniform method for the entire scene. Chen wants to adapt the algorithm to the content of the scene. Parts that contain high frequency content, i.e., where there is a lot of change in the scene, will take more processing than those parts with smoother content. Graduate student Che-Hung Kuo is working with Professor Chen to adjust the anti-aliasing method to adapt to the frequency of the content.

For more information, please see Chen and Interrante's Web pages.

-Bobbie Othmer