Title: "Computational Algorithms for Image Registration and Shape Recovery."

Speaker: Baba Vemuri
         Dept. of CISE
	 Univ. of Florida
	 Gainesville, Fl. 32611			


			  Abstract

Image registration and 3D shape recovery are fundamental problems in
computer vision with applications in Medical Imaging and various other
fields of Engineering and Science. In this talk, for the first half, I
will focus on the problem of image registration and present a robust
and fast registration algorithm and demonstrate its application via
examples in the field of medical imaging specifically to pre and
post-operative MRI registration.  For the second half, I will present
a hybrid (rigid+non-rigid) modeling scheme and the associated
numerical algorithm for model fitting to 3D data.

Estimating the registration between two volume data sets is formulated
as a motion estimation problem. A hierarchical optical flow motion
model is used that allows for both global as well as local motion
between the data sets.  In this model, the flow field is represented
by a B-spline basis which implicitly incorporates smoothness
constraints on the field. Motion between the data sets is estimated by
minimizing the expectation of the squared differences energy function
numerically via a modified quasi-Newton iteration scheme.  The key
idea in the modified Newton method is that the Hessian of the
objective function at the optimum is pre-computed.  I will present
examples demonstrating the performance of the algorithm on synthesized
and real data.

The shape recovery problem is formulated in an active (geometric)
modeling framework. A novel modeling scheme is introduced which
consists of representing shapes by pedal curves and surfaces -- pedal
curves/surfaces are the loci of the foot of perpendiculars to the
tangents of a fixed curve/surface from a fixed point called the pedal
point. By varying the location of the pedal point, one can synthesize
a large class of shapes which exhibit both local and global
deformations. We introduce physics-based control for shaping these
geometric models by letting the pedal point vary and use a dynamic
spline a.k.a. ``snake'' to represent the position of this varying
pedal point. The model dubbed as a ``snake pedal'' allows for
interactive manipulation as well as topological changes.  I will show
examples of model fitting to shapes of interest in 3D range and volume
data.