Delphine Nain

Research

PhD Projects @ Georgia Tech • Masters @ MIT, AI LabUndergraduate @ MIT, Media Lab

PhD Projects @ Georgia Tech

In my PhD work at Georgia Tech I am co-advised by Professors Allen Tannenbaum and Aaron Bobick and a member of the MINERVA group and the CPL lab. My thesis focus is shape analysis and segmentation. Below are some of the projects I've worked on since starting my PhD.

multiscale 3D shape analysis

Multiscale 3D Shape Analysis using Spherical
Wavelets

with Steven Haker, Aaron Bobick and Allen Tannenbaum

Shape priors attempt to represent biological variations within a population. We present a novel algorithm that learns shape variations from data at multiple scales and locations using spherical wavelets and spectral graph partitioning. Our results show that for a given training set size, our algorithm significantly improves the approximation of shapes in a testing set over PCA, which tends to oversmooth data.

     
vessel 1 vessel 2

Vessel Segmentation Using a Shape Driven Flow
with Tony Yezzi and Greg Turk.

We present a segmentation method for vessels using an implicit deformable model with a soft shape prior. We combine image statistics and shape information to derive a region-based active contour that segments tubular structures and penalizes leakages. We present results on synthetic and real 2D and 3D datasets.

   

Algorithms for Stochastic Approximations of Curvature Flows
with Gozde Unal and Allen Tannenbaum

 

geometric heat equation, stochastic curve evolution

     
dimes 0001
dimes 0500

Level Set methods for Curve Evolution
and Image segmentation

with Tony Yezzi and Greg Turk.

     
bunny 1
bunny 2

Spectral Analysis of Mesh Geometry
and its application to mesh subdivisions.

(3 week mini-project for CS7001),
with Greg Turk.

     

Matching Cartoon Characters using
Active Shape Model Techniques.

(A three week mini-project for CS7001),
with Frank Dellaert.

 

Masters @ MIT, AI Lab

I completed my Master's Thesis in May 2002 under the guidance of Professor Grimson in the Medical Vision Group at the AI Lab. Specifically, I worked on automatic path planning algorithm for a virtual endoscopy tool that I developed as part of my bachelor's project. For this project, I worked in collaboration with researchers at the Surgical Planning Laboratory of the Harvard Medical School at the Brigham and Women's hospital in Boston. If you want to learn more, check out the paper that I presented at the IMIVA workshop of the medical imaging conference MICCAI 2001 in Utrecht.

Here is also a link to an article in Virtual Medical Worlds about the workshop and each presentation.

 

Undergraduate @ MIT, Media Lab

As an undergraduate I had a lot of fun working at the Media Lab, in theSynthetic Characters Group (1998-2000). I helped developed an autonomous camera system in interactive 3D installations, and presented one of our installations
(Void *): A Cast Of Characters at the Computer Graphics conference SIGGRAPH'99 in LA with the rest of the team.

To learn more, check out this paper that Bill Tomlinson and I presented at the Autonomous Agents 2000 conference in Barcelona.