My primary research focus is in visualization - designing and implementing algorithms for portraying large, complicated datasets in ways that facilitate the rapid, intuitive appreciation of the essential features of their contents. In the design component of my work I try to integrate inspiration from art and illustration with insights from research in human vision and perception to elucidate an objective understanding of how to define effective techniques for visually communicating the essential contents of a dataset. For example, I am currently working on a project whose goal is to investigate the effect of texture pattern anisotropy on surface shape perception; the purpose in this effort is to attain a understanding of how to select (or custom-define) and apply a texture pattern to a smoothly curving surface in order to enable a more accurate and intuitive understanding of its 3D shape than might be possible either in the absence of texture or if a standard alternative texturing method were used. Of course all of my work requires significant implementation efforts, and this component of the work is where I often must draw from a broad range of disciplines including applied mathematics, computer vision, image processing and computer graphics to define the algorithms that are necessary to achieve the results that I desire.

For example, one of the hypotheses in the texturing project that I mentioned above is that the orientation of an anisotropic pattern with respect to the lines of maximum and minimum normal curvature over the surface of an object may affect an observer's perception of the underlying surface shape. If this is true, then it means that there are better and worse ways to apply a given texture pattern to a surface in order to emphasize its shape. In order to test this hypothesis it is first necessary to be able to compute the principal directions everywhere over a surface and then to synthesize an anisotropic texture pattern over the surface such that the direction of the anisotropy either follows the principal directions or does not. If the hypothesis is true, in order for the results to be of practical benefit for applications involving acquired data it is necessary to be able to accurately estimate the principal direction and lines of curvature over arbitrary, polygonally-defined meshes, and to be able to synthesize a wide variety of useful and interesting textures (ideally from a provided 2D example). With the help of my students and collaborators, I am currently working on all of these different aspects of this shape-representation-with-texture project.

A second project that I am currently working on involves the design, synthesis and use of texture patterns for multivariate data visualization. The essence of this idea is that if we can define a set of independent perceptual dimensions for texture, we may be able to synthesize a multidimensional texture palette in which observers have the ability to accurately and intuitively identify the position of any individual pattern in the distribution. This collection of texture patterns could then potentially be used to effectively communicate a vector of values at all points in a 2D map. There are again multiple components to this project: we have to verify that human observers can reliably and independently perceive the individual texture features that are traditionally thought to comprise the perceptual dimensions of texture, then we must determine how to identify and how to select or synthesize a perceptually equidistant, multidimensional texture palette, and finally we must realize the system and test the results for effectiveness.

In addition to these two projects that have a significant component involving perception, I am working on several other projects in which the main emphasis is in feature extraction for visualization. These include a joint effort with aerospace and mechanical engineering faculty to automatically segment and track the motion, evolution and interactions of possibly densely packed individual vortices in a 3D turbulent boundary layer, and a joint effort with researchers in child development and ophthalmology that involves determining information about the gaze direction of infants from uncalibrated photographs, which I mention for completeness in communicating my interests and background.

There are several things that I would like to learn from this campfire. I am looking forward to the opportunity to meet other people who are working on problems that involve perception and graphics and to find out more about what everyone else is doing. I also hope to gain insight into what the most important directions are for future work in perception-based visual representation, and how to more effectively develop productive collaborative relationships that can advance these efforts.

Other potential discussion questions that interest me are: how can we best mine the store of existing knowledge in visual perception for guidance in visualization design - how can we make the leap from what psychologists have learned about vision and attention through simple controlled experiments to the design of techniques for communicating complicated information in a more effective way? Also, to what extent do we need to and how might we more effectively assess the benefits of new visualization techniques? The current culture in our field is that experimental evaluations are not required for publication; what are the obstacles in conducting effective experiments, and how might we facilitate the dissemination of information or software that could assist more people in experimental design and the analysis of results?

In conclusion, my philosophy of research in visualization is that there is a science behind the art of effective visual representation, an explanation for why some representations are more effective than others. The more clearly we understand how our visual system works, the better equipped we are to be able to wisely choose appropriate directions for our efforts in designing effective approaches to the visual communication of information. However, design is much more than the straightforward application of principles, and I believe that insights for visualization from perception can be enriched by inspiration from the works of artists and illustrators, who have a long history of experience and their own unique insights into perception and visual representation.