Professor Thompson's primary research interest is in the area of
computational vision, with an emphasis on problems involving the
determination of spatial organization.
This research focuses on increasing the effectiveness of computer
graphics in conveying information about the three-dimensional world. It
is an interdisciplinary effort involving computer graphics,
perceptual psychology, and computational vision. We are interested in
better understanding the spatial information potentially available in
CG imagery, determining what spatial cues are actually used when CG
imagery is viewed, and using this information to inform the development
of improved rendering algorithms.
Research ultimately aimed at provide tools to enable the design of safe
environments for the mobility of low-vision individuals who may need to
operate under low luminance and other visually challenging
conditions. This is a multi-university, multi-disciplinary effort
involving clinical low vision, visual science, perceptual psychology,
and computer graphics.
Much of Prof. Thompson's prior research deal with the analysis of
visual motion. This included optical flow estimation, detection and
interpretation of discontinuities in optical flow, and the detection of
moving objects from non-stationary viewpoints.
A model of the process by which current location can be determined
from views of outdoor environments and maps representing topographic
information was developed. This required an interdisciplinary approach,
involving laboratory and field studies of highly expert map readers,
computational analysis, and computer simulations.
Accurately rendering naturally occurring outdoor terrain requires a
number of techniques that are very different from those involved in
realistic rendering of indoor scenes. Prof. Thompson's work has
included methods for constructing models of terrain features that
combine image and elevation data to produce a model more detailed than
would be possible with elevation data alone, plus methods for
augmenting terrain models with correct vegetation and snow coverage.