BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Frameworks for segmentation\, classification\, and nonstationary i
 mage processing with applications to disease assessment - Albert Montillo\
 , University of Pennsylvania
DTSTART:20100706T130000Z
DTEND:20100706T140000Z
UID:TALK25404@talks.cam.ac.uk
CONTACT:Microsoft Research Cambridge Talks Admins
DESCRIPTION:*Abstract:* This talk presents several frameworks for image an
 alysis that solve ubiquitous challenges for computer vision. The first fra
 mework addresses the dynamic segmentation of 3D deformable objects. For ro
 bust segmentation of objects undergoing complex motion\, a key step is the
  proper implementation of conservation laws that govern shape deformation.
  I show how to reconstruct 3D geometry using a model from data-fusion appr
 oach and how to implement physically sound\, compressibility constraints f
 or real materials. The concepts are generally applicable to segment deform
 able shapes imaged over time\, and I illustrate them to segment the heart 
 in cardiac images. The second framework targets lower level image processi
 ng challenges including: denoising and motion tracking. I recast these pro
 blems in a nonstationary setting and show which properties vary over space
  and how to utilize that information to optimize image processing results.
  For denoising\, I show how to construct a new\, generic\, nonstationary d
 enoising framework that preserves more true structure for better image int
 erpretation. For motion tracking\, I show how the motion of objects with c
 haracteristic local line patterns can be best recovered by nonstationary w
 eighting of Fourier domain information based on local content in the spati
 al domain. The third image analysis framework addresses probabilistic clas
 sification. This framework applies to problems where the primary geometric
  constraint is difficult to express concisely with a conservation law\, bu
 t can be recovered probabilistically by learning from training data. The f
 ramework entails several steps including: data registration to form an atl
 as space\, feature extraction\, and probabilistic model construction. I wi
 ll demonstrate the framework with two different approaches for the model c
 onstruction step. First\, using a Bayesian learning approach with a Markov
  Random Field\, I will show how the framework enables a 10 fold increase i
 n the number of structures that can be automatically labeled in brain MRI.
  The method has received FDA approval for neuroanatomical structure quanti
 fication and its applications include disease assessment of: Alzheimer`s\,
  Huntington`s\, Parkinson`s and Schizophrenia. Second\, using a discrimina
 tive learning approach with the Random Forest\, I show how the framework c
 an be applied to extract biometric data about a person from an image of th
 eir face. Applications include security and surveillance tasks. 
LOCATION:Small Lecture Room\, Microsoft Research\, Roger Needham Building\
 , 7 J J Thomson Avenue\, Cambridge CB3 0FB
END:VEVENT
END:VCALENDAR