SPEAKER: KATSUSHI IKEUCHI
Professor, Institute of Industrial Science, University of Tokyo
ABSTRACT:
SPEAKER BIO:
Dr. Ikeuchi developed the "smoothness constraint," a constraint to
force neighboring points to have similar surface orientations; this
constraint enabled him to iteratively recover shape from shading and
shape from texture. He pioneered the use of specular reflections to
recover surface orientations. Instead of discarding specular
reflections, he effectively used them for recovering shape and
reflectance. Recently, this method evolved into Photometric Sampling,
which determines not only the object's shape but also its surface
characteristics. He has also been working to develop object
representations for vision-guided manipulation for such tasks as
bin-picking of man-made objects, sampling of natural objects for
planetary rovers, or clean-up of nuclear accident sites. The
representations he has proposed include the Extended Gaussian Image
(EGI), the Complex Extended Gaussian Image (CEGI), the Spherical Angle
Image (SAI), and a frame-based geometric/sensor modeling system
(VANTAGE).
Recently, Dr. Ikeuchi's main focus has been the development of vision
techniques that enable a reduction in programming efforts. These
techniques include: 1) modeling-from-reality, which automatically
acquires geometric and photometric models of objects by simply
observing actual objects, 2) vision-algorithm-compiler, which
automatically converts object and sensor models into recognition
programs, and 3) assembly-plan-from-observation, which, by observing
human assembly actions, acquires robot assembly programs that mimic
those actions.
Conferences for which Dr. Ikeuchi has served as general/program-chair
include the 1995 Intelligent Robotocs amd Systems (IROS) Conference
and the 1996 International Conference on Computer Vision and Pattern
Recognition (CVPR). He has also served on the program committees of
several international conferences. He is on the editorial board of
the IEEE Transactions on Pattern Analysis and Machine Intelligence,
the International Journal of Computer Vision, the Journal of Computer
Vision, Graphics, and Image Processing, and the Journal of Optical
Society of America. He is a fellow of IEEE.
He has received several awards, including the David Marr Prize in
computational vision, and an IEEE outstanding paper award. In addition, in
1992, his paper, "Numerical Shape from Shading and Occluding Boundaries,"
was selected as one of the most influential papers to have appeared in
Artificial Intelligence Journal within the past ten years.
Modeling from Reality
One of the promising applications of computer vision is modeling-
from-reality. This application creates a model for a virtual reality
system by simply observing a real environment/object and automatically
creating models for it by using computer vision techniques. Such a
method drastically reduces the cost of a virtual reality system and
opens up a wider application of that system. Along with Raj, I began
this effort to create such systems in 1990. We have since developed
methods to recover both geometric and photometric models from
observation. The first half of my talk is devoted to a quick overview
of techniques for recovering geometric models. The second half of my
talk focusses on techniques for recovering photometric models. One of
those techniques deals with recovering reflectance properties of an
object from a sequence of color and range images. I will also present
a method for integrating such a computer-generated model with a real
scene in order to produce a mixed reality system.
Dr. Katsushi Ikeuchi is a Professor at the Institute of Industrial
Science, the University of Tokyo, Tokyo, Japan. He received the B.
Eng. degree in Mechanical Engineering from Kyoto University, Kyoto,
Japan, in 1973, and the Ph.D. degree in Information Engineering from
the University of Tokyo, Tokyo, Japan, in 1978. After working at the
Artificial Intelligence Laboratory at Massachusetts Institute of
Technology, the Electrotechnical Laboratory of the Ministry of
International Trade and Industries, and the School of Computer
Science, Carnegie Mellon University, he joined the University of
Tokyo, in 1996.