David Breen's team is developing a novel volumetric three dimensional
morphing technique, in conjunction with Dr. Ross Whitaker from the University
of Tennessee, Knoxville. First, Dr. Whitaker's MRI image data segmentation
technique is used to output a clean discrete 3-D volumetric image data. His
technique removes false objects created by noise, and then in principle wraps
an imaginary flexible sphere around the object to produce a smooth surface
surrounding the object to be morphed. I would integrate his work into David
Breen's process to produce cleaner and more accurate results.
The second component of the technique relies on a distance measure in a 3-D
image space which measures the distance between a given point in space to the
nearest point on the object being morphed. Sean Mauch developed a mathematical
solution which essentially propagates a distance measurement solution given
around the boundary of the image object out into space. For instance, given a
3-D image of a face and the solution of the distance measure around the skin
and hair, Sean Mauch's solution propagates these values outwards, i.e. away
from the face. Therefore, the major implementation goal of my research would
be to efficiently compute this distance measure at the boundary of the image
object. The brute force technique used currently involves computing the
distance measure at each point in space by slowly growing an imaginary sphere
around that point until it hits the image object, and taking the radius of that
sphere to be the distance measure. Unfortunately, this technique is too slow
to be acceptable given the size and complexity of the typical MRI image data.
Finally, the above results would be input to the general morphing code. In
conclusion, I would like to complete an animation of this completed process by
the October 1998 deadline for a journal publication.
California Institute of Technology: Computer Science: Graphics Group: David Breen, Sean Mauch, Yaniv Inbar.|
University of Tennessee: Department of Electrical Engineering: Ross Whitaker.
Sean's A Fast Marching Method of Computing Closest Points (J A Sethian)
Dave's 3D Scan Conversion of CSG Models Into Distance Volumes