G.Ünal; "Variational Techniques in Problems...", Dec. 3, 13:40, L030
  • FENS
  • G.Ünal; "Variational Techniques in Problems...", Dec. 3, 13:40, L030

You are here

Faculty of Engineering and Natural Sciences








Variational Techniques in Problems of Image Segmentation,


Registration, and 3D Reconstruction



Gözde Ünal


Sabancı University





Recently, tools of variational calculus have found extensive applications in computer vision.


In this talk, I will present variational approaches to problems in image segmentation and registration. I will show various deformable models and their applications in medical image segmentation problems for lymph nodes, vertebrae, vessels and such. I will also talk about the coupled partial differential equations (PDEs) we proposed to estimate vector fields that define the deformation between objects, and the surfaces that define the segmentation of the objects simultaneously. Finally, I will briefly talk about PDE-based 3D object reconstruction methods.




Short Bio:


Gözde Ünal received her PhD degree in electrical and computer engineering from North Carolina State University in 2002, and then was a postdoctoral fellow at the Georgia Institute of Technology for a year. Between 2003 and 2007, she was a research scientist at Siemens Corporate Research, Princeton, NJ, USA. She has been a faculty member at Sabancı University since Fall 2007. Her current research is focused on medical image analysis, segmentation, registration, and shape analysis techniques with applications to clinically relevant problems in MR, CT, US, and intravascular images.





Dec. 3, 2008, 13:40, FENS L030