Title: Applications of 2D and 3D Differential Optical Flow
Speaker: Prof. John Barron, Dept. of Computer Science,
Univ. of Western Ontario, London, Ontario,
Canada (Currently on sabbatical with the
Vision Systems Group, DCU)
Venue: Room S209, Research and Engineering Building,
Dublin City University
About the Speaker
John Barron was born in Corner Brook, Newfoundland, Canada. He graduated with Physics and Computer Science degrees from Memorial University in St. John’s, Newfoundland and obtained his MSc and PhD degrees from the University of Toronto in 1988.
He is current a professor in the Computer Science department at the
University of Western Ontario in London, Ontario. His research interests
are in Image Processing and Computer Vision, in particular, in the
measurement and interpretation of 2D/3D optical flow.
About 2D/3D Optical Flow
2D optical flow is an approximation to the local image motion in a
sequence of images. When 2D optical flow is introduced, the 2D
Motion Constraint Equation results in what is called the 2D aperture problem. Two common optical flow algorithms are then described to overcome this problem using optical flow computations. Then two applications of 2D optical flow are shown: the recovery of camera motion and scene depth from time-varying optical flow and the measurement of 2D/3D corn seedling motion and growth via optical flow.
3D optical flow is an approximation to the local volumetric motion is a
sequence of volume images. Closely related to it is 3D range flow, which
is an approximation to the local surface motion in a sequence of depth
images. 3D optical flow is first introduced, then there is a demonstration of how 3D Motion Constraint Equation results in the 3D aperture problem. 3D extensions of the 2 algorithms described above overcome this problem. Finally, 3 applications of 3D optical flow are desribed: the use of 3D range flow to measure the motion and growth of plant leaves, the use of 3D optical flow to measure/predict Doppler Radar storm motion and the use of 3D optical flow to measure the motion of a beating heart in gated MRI datasets. The last project is ongoing research with Prof. Paul Whelan at DCU.
This talk is non-technical and equations have been kept to a minimum.