"Analysis of Medical Images Based On Physics Models"
Prof. Michael Brady
BP Professor of Information Engineering
Oxford University, UK
Medical Vision Laboratory Web Page
Prof. Michael Brady's Web Page
We describe recent work on the detection of breast cancer using x-ray and
contrast-enhanced MR mammography. We show that in order to build robust,
reliable programs that can be used routinely in the clinic, it is necessary to
develop physics-based models of image formation. We demonstrate very
promising results in both cases.
"An Expectation-based, Multi-focal Saccadic Vision System for vehicle guidance - (EMS-Vision)"
Prof. Ernst D. Dickmanns
ISF, Federal Armed Forces University, Munich (UniBwM)
Prof. Ernst D. Dickmanns' Web Page
Based on a decade of experience with bifocal vision systems in
real traffic situations the new system has been designed. It consists of at
least three (better four) cameras mounted fixed relative to each other on a
highly dynamic pan and tilt platform with the capability of inertial
stabilization, smooth pursuit and fast saccades. The overall system design,
its realization with off-the-shelf PC hardware and its applications to road
vehicle guidance will be discussed.
"Development of Vision Guided Autonomous Systems: Interaction of Real-Time Vision and Control"
Prof. Takeo Kanade
U. A. and Helen Whitaker Professor of Computer Science
Director of the Robotics Institute
Carnegie Mellon University
Prof. Takeo Kanade's Web Page
Development of real-time vision systems for controlling fast autonomous
systems presents challenges not only for developing capable vision
algorithms but also for making them work in concert with control of a
physical system which has its own dynamics. This class of system (for
I coin the phrase a real real-time system) requires special care with
regard to latency, bandwidth, reliability, system dynamics model, and
fusion of multi-modal sensors.
I will discuss the two such systems developed at the Robotics Institute
Carnegie Mellon University: Navlab and Robocopter. The CMU Navlab
Project has a long history of development, and so far ten systems have been
most recently, many advanced capabilities, including lane following,
obstacle detection by sensor fusion, lane change, blind-spot detection,
tracking a car in front of the vehicle, have been integrated into a
system. The CMU Robocopter, equipped with multiple cameras, laser
GPS, gyros, INS, and multi-DSP computers, can now stably take off, land,
and free-fly while observing the ground below and tracking objects by a
real-time vision system.